AI’s Energy Appetite Is Outpacing Deployment of AI-Based Climate Solutions: IEA

Why It Matters

The widening gap threatens higher emissions and energy costs, undermining AI’s promise to accelerate the clean‑energy transition and raising ESG risks for investors and policymakers.

Key Takeaways

• AI data‑centre electricity demand could reach 950 TWh by 2030.
• Energy‑sector AI adoption lagging due to skills, data, and policy gaps.
• Tech firms’ capex exceeds $400 bn, outspending global oil‑gas investment.
• IEA projects AI could cut 13 exajoules of energy by 2035.

Pulse Analysis

The surge in AI‑driven workloads is reshaping the global data‑centre landscape. Southeast Asia has emerged as a critical hub, with "AI factories" tripling in capacity over the past 18 months. The IEA projects that electricity consumption by AI‑focused facilities will climb from 485 TWh in 2025 to 950 TWh by 2030—about 3 % of worldwide demand—and that a single advanced server rack could demand as much power as 65 households by 2027. This rapid expansion is fueled by capital expenditures exceeding $400 billion in 2025, a level that now eclipses total investment in oil and gas production.

Despite the mounting demand, the energy sector’s own use of AI remains sluggish. A survey cited by the IEA highlights chronic shortages of digital talent, fragmented data ecosystems, cybersecurity concerns, and weak policy incentives as primary barriers. Yet AI already demonstrates tangible benefits: it can monitor transformers, optimise industrial processes, and improve grid reliability. If scaled, existing AI applications could shave more than 13 exajoules—roughly 3 % of global final energy consumption—off the demand curve by 2035, delivering 3‑10 percentage‑point cost reductions in energy‑intensive industries.

For investors and policymakers, the report signals a dual imperative. On one hand, unchecked data‑centre growth risks higher electricity prices, strained grid infrastructure, and heightened ESG scrutiny. On the other, unlocking AI’s efficiency potential could turn data centres into flexible grid assets, especially as they pair with 20‑25 GW of battery storage projected for 2030. Coordinated policy frameworks, workforce upskilling, and robust cybersecurity standards are essential to bridge the adoption gap and ensure AI serves as a catalyst rather than a burden for the energy transition.