2025 in review – AI and sustainability

In 2025, the relationship between artificial intelligence and sustainability moved from hypothetical to unavoidable. What had once been framed as a promising alignment – AI as an enabler of climate action, efficiency, and smarter systems – became a more contested, nuanced debate. 

Across the year, a clear picture emerged: AI’s sustainability impact is neither inherently positive nor negative, but deeply shaped by governance, energy systems, and corporate choices.

The energy reality check

Few issues dominated the conversation more than energy demand. Warnings that data centres could consume up to 12% of US electricity within just a few years sharpened concerns that AI’s growth trajectory may collide with climate goals. Subsequent analysis suggested that surging data centre demand could actively delay fossil fuel phase-outs in some regions, particularly where grids remain carbon-intensive.

These concerns filtered quickly into boardrooms. Surveys showed that energy use and emissions uncertainty began dampening business enthusiasm for deploying AI, specifically for sustainability outcomes. Rather than being viewed as a climate accelerant, AI increasingly looked like a trade-off – at least in the short term.

High-profile scrutiny, including warnings from former Microsoft employees on the use of AI to maximise oil and gas extraction, underscored internal tensions between growth ambitions and sustainability responsibilities.

Measuring what matters – or failing to

A recurring theme throughout 2025 was how little is still known about AI’s true environmental footprint. Efforts to catalogue the emissions associated with AI queries, model training, and deployment revealed a fragmented and inconsistent measurement landscape. Even comprehensive explainers concluded that major data gaps persist, making confident claims difficult.

Technology companies responded with increased transparency, publishing more disclosures around AI-related emissions. Yet reporting often outpaced action, with critics noting that clarity on impacts was rarely matched by binding reduction commitments. 

AI as a climate solution – with conditions

Despite the concerns, the case for AI as a powerful climate tool remained compelling. The International Energy Agency argued that AI could enable emissions reductions up to three times greater than its own footprint, particularly in energy systems, transport, and industry – if deployed strategically.

This conditional optimism became a defining tone of the year. Rather than asking whether AI is “good” or “bad” for sustainability, the conversation shifted to how it is applied, and under what governance frameworks. Without alignment to clean energy, efficiency gains risk being overwhelmed by rebound effects and scale.

Governance gaps and ESG risk

Another stark finding of 2025 was how unprepared many organisations remain for the age of AI. Research suggested that just 2% of firms have adequate responsible AI measures in place, raising red flags for sustainability, ethics, and long-term value creation. At the same time, companies were deploying AI at unprecedented speed, often without fully considering ESG risks or downstream impacts.

For Chief Sustainability Officers, this created both pressure and opportunity. AI emerged as a force reshaping the CSO role itself – demanding greater technical literacy, closer collaboration with IT and data teams, and more assertive governance influence.

Read also: AI skills every sustainability leader should learn

Collaboration, benchmarks, and coalitions

With more people recognising the scale of the challenge, 2025 also saw a surge in collaborative initiatives. The World Economic Forum outlined a roadmap for sustainable AI, calling for shared standards and cross-sector cooperation. Governments and private sector actors also launched new coalitions focused on aligning AI development with environmental goals.

On the corporate side, tools such as Salesforce’s AI energy efficiency benchmarking platform pointed toward a more data-driven approach to managing impacts. At the global level, the UN’s adoption of its first resolution on AI and environmental sustainability marked an important symbolic step, even as questions remained about enforcement and implementation.

A shift in narrative

Perhaps most telling was a subtle shift in corporate climate communications. Moves such as Google stepping away from explicit net-zero language signalled a broader recalibration – an acknowledgement that AI-driven growth is complicating traditional sustainability narratives.

By the end of 2025, the message was clear: AI will shape the sustainability agenda whether organisations are ready or not. The year did not deliver simple answers, but it did establish a new baseline of realism. AI’s environmental impact is now a material sustainability issue – one that demands measurement, governance and collaboration.