Artificial intelligence has moved from experiment to infrastructure in financial services. That scale brings a resource footprint that can clash with regulators’ and investors’ expectations for ESG and Net Zero performance. Finance and insurance leaders must treat AI as a material source of operational and transition risk.
The Resource Intensity of Artificial Intelligence
Energy Demands: Every AI Query Adds Up
Estimates indicate a single large-language-model query can consume five to ten times the energy of a conventional web search, roughly the equivalent of a 60 watt lightbulb running for about 20 minutes. With prompt volumes now measured in the billions daily, aggregate energy use for inference and training is rising quickly. Data centers currently account for roughly 1 percent of global electricity; AI workloads are a growing share of that demand.
Water Consumption: Cooling the Digital Brains
Cooling high-density compute is water intensive. A commonly cited figure is about 500 millilitres of water per 100 words generated by some large models. Scaled across millions of daily requests, that adds up to substantial freshwater withdrawals, particularly in regions that rely on evaporative cooling or non-recycled water supplies.
ESG Imperatives and the Financial Sector
Insurers and financial firms in the UK, EU and US face tightening disclosure and operational rules, from climate-related reporting requirements to sector-specific sustainability guidance. Environmental impacts tied to AI procurement, hosting and model training can create a compliance gap if not measured and reported. In addition, reputational risk and underpriced transition exposures emerge when technology-driven emissions are ignored in credit, underwriting and investment decisions.
Towards Sustainable AI Solutions and Outlook
Governments and industry are responding with energy efficiency targets, standards for model transparency, and incentives for low-carbon compute. Technical approaches include model compression, specialised chips with better performance per watt, workload scheduling to match renewable availability, and water-efficient cooling. Cloud providers are also offering regions powered by renewables and options for recycled water cooling.
Implications for FinanceAIInsiders Readers
For finance and insurance executives, AI is now both an opportunity and a risk vector. Practical steps include quantifying lifecycle emissions for AI assets, adding AI-related metrics to ESG reporting, setting procurement criteria for low-carbon compute, and reflecting technology-driven transition risk in underwriting and investment models. Early measurement and governance will reduce regulatory exposure and protect franchise value as AI use scales.
