AI's Power Hungry Data Centres to Exceed 190 Nations’ Energy Use by 2030: UN Transparency Push

Artificial intelligence is rapidly becoming one of the most significant unregulated drivers of global greenhouse gas emissions and water scarcity. For climate and energy professionals, the UN’s new AI Environmental Transparency Initiative represents a pivotal moment: if major tech firms truly reveal the carbon, water, and land toll of their data centres, it could close a massive gap in corporate climate accountability and set a precedent for hard-to-decarbonise sectors.

United Nations Secretary-General António Guterres sharply escalated the global debate over artificial intelligence’s environmental toll on June 23, 2026, using his platform at London Climate Action Week to demand that major AI firms fully disclose the water, carbon, and land-use impacts of their data centres—and commit to powering them entirely with renewable energy by 2030. The newly launched UN AI Environmental Transparency Initiative marks the first coordinated international push to bring the opaque, resource-hungry AI infrastructure sector under public scrutiny, as the world simultaneously grapples with worsening climate indicators and a crisis of corporate net-zero credibility. Guterres presented stark projections: if current trends continue, the world’s AI data centres could by 2030 consume more electricity than all but the five largest national energy users, while their water withdrawals would be sufficient to meet the basic annual needs of all 1.3 billion people in sub-Saharan Africa.

The timing of the UN call reflects a rapid shift in AI’s physical footprint. The compute requirements for training frontier models are doubling roughly every six months, and hyperscale data centre construction has become the single largest infrastructure capex category for leading cloud providers. Yet, despite a proliferation of voluntary sustainability pledges from the major technology companies—Microsoft’s carbon-negative commitment, Google’s 24/7 carbon-free energy pledge, Amazon’s 2040 net-zero target—actual transparency remains limited. Many firms do not publicly disaggregate water consumption per facility, do not report location-specific land-use change, and increasingly rely on power purchase agreements that offset rather than directly fuel operations with renewables. Guterres explicitly criticised this voluntary framework as insufficient, framing mandatory transparency as the necessary precursor to meaningful accountability.

The secretary-general explicitly linked the AI transparency push to broader fossil fuel phase-out demands, taking aim at the global methane crisis in parallel. He launched a separate call to action requiring fossil fuel producers to fix leaks, stop routine flaring, and adopt a uniform science-based global methane measurement standard. With methane responsible for roughly one-third of current warming, the twin initiatives are designed to squeeze both the demand- and supply-side drivers of emissions—the emerging AI economy on one flank, legacy oil and gas operations on the other. Industry observers note that this dual pressure could accelerate regulatory convergence between climate and technology policy, which have historically been treated as separate domains.

From a market and compliance perspective, the implications are significant. If the UN initiative gains traction, AI service providers—including hyperscalers, co-location operators, and chip manufacturers—will soon face a layered patchwork of disclosure mandates, building on the EU’s Corporate Sustainability Reporting Directive (CSRD) and California’s climate disclosure laws. The reported data could directly influence enterprise procurement decisions, as corporate buyers of cloud AI services increasingly factor Scope 3 emissions into their own regulatory filings. For investors, enhanced transparency exposes water-risk liabilities in drought-prone regions where many data centres are concentrated, potentially re-rating asset valuations.