A new survey by data centre provider Telehouse Europe reveals that more than half of UK IT decision-makers admit their organizations cannot accurately measure the carbon emissions generated by AI workloads, despite widespread optimism about AI’s role in achieving net-zero and broader ESG goals.
The research found that 56% of respondents struggle with reliably quantifying the environmental impact of AI operations, creating a significant barrier for companies aiming to demonstrate tangible sustainability outcomes. While 89% of UK IT leaders believe AI accelerates progress toward their net-zero and ESG targets, the lack of precise emissions data means many cannot confidently link AI use to measurable environmental benefits.
The survey highlighted a disconnect between organizations’ strategic ambitions and their immediate operational focus. About 45% of respondents prioritize sustainability in their AI-related spending, yet 62% primarily focus on productivity improvements and 49% on cost savings. This suggests that while sustainability is recognized as a key long-term goal, short-term operational metrics often dominate budget allocations.
Looking ahead, 79% of IT decision-makers see AI as vital to meeting their 2030 sustainability targets. However, only 23% consider renewable energy sourcing and efficiency when selecting AI infrastructure providers, indicating that energy sourcing criteria are currently underrepresented in procurement decisions despite the high electricity demands associated with training and deploying AI models.
The report emphasizes the governance challenges surrounding AI’s environmental footprint, especially as AI workloads often span multiple data centres, cloud providers, and shared infrastructure. Inconsistent reporting methods and the absence of standardized internal metrics hinder organizations’ ability to track emissions effectively.
Additionally, the survey reveals that 93% of decision-makers are influenced by geopolitical uncertainties when planning their infrastructure strategies. Yet, 59% estimate it would take over an hour to switch AI inference workloads to alternative sites in the event of a regional incident, highlighting resilience vulnerabilities as reliance on AI increases.
Mark Pestridge, Executive Vice President and General Manager of Telehouse Europe, noted the gap between confidence in AI’s sustainability benefits and the ability to measure its environmental impact. “AI has earned its place in sustainability strategies, but belief in its potential is running ahead of evidence. Until carbon visibility improves and sustainability becomes a core criterion in infrastructure decisions, the sector’s green ambitions will remain more aspiration than proof,” he said.
The findings underscore ongoing debates around how organizations account for emissions from digital services like AI and incorporate that data into ESG reporting and procurement. As AI becomes a mainstream technology investment, aligning its deployment with mature carbon accounting practices will be critical to turning sustainability goals into measurable results. Telehouse Europe emphasizes that addressing these measurement and resilience challenges will be vital as AI continues to expand across customer services, analytics, and automation.
