From Total Cost of Ownership to Social Impact: A Frugal AI Framework to Measure Your AI Portfolio as a Strategic Asset

The comprehensive measurement of the value of an Artificial Intelligence (AI) portfolio remains a significant challenge within organisations. Assessments are often nascent and fragmented, failing to account for the full spectrum of costs and impacts. This paper proposes a standardised, three-level framework for measuring AI value, developed in partnership between the Frugal AI Hub at the University of Cambridge and the United Nations International Computing Centre (UNICC). The framework’s metrics are designed for both the individual AI model and the overall portfolio in the context of an organisation or entity ’users’ of AI. Level 1 focusses on the Total Cost of Ownership (TCO) and the Frugal AI metrics to quantify financial and energy costs. Level 2 introduces a return-on-investment (ROI) measurement that links financial benefits to TCO. Finally, Level 3 aligns AI performance with the Sustainable Development Goals (SDGs) of the United Nations to measure social impact. By integrating these metrics, the framework enables organisations to move beyond siloed assessments, optimising their AI portfolios for efficiency and responsible deployment. Although developed in conjunction with the UNICC, this framework is broadly applicable to any organisation seeking a transparent and equitable valuation of AI. The framework draws on a UNICC survey of two multi-agency AI use cases covering the full design-to-run lifecycle costs, which inform the cost component breakdowns and validation of regression models. This paper is the first of a wider research program. Other papers in progress or in the pipeline focus on further refinements of the metrics, examples of operationalisation and portfolio optimisation.