Aon Executive Proposes AI FinOps Framework for Scaling GenAI
Sudipto Dasgupta of Aon advocates for an AI FinOps framework and centralized control plane to manage the costs and operational complexities of generative AI.
Sudipto Dasgupta, Global Head of Data, AI, and Automation Platforms at Aon, is calling for the adoption of an AI FinOps framework to address the financial and operational hurdles of scaling generative AI. He argues that traditional FinOps practices cannot handle the specific demands of AI, such as variable token-based pricing, the complexity of multi-model environments, and the rapid evolution of foundation models.
To prevent budget overruns and manage costs linked to agentic workflows, Dasgupta proposes the implementation of a centralized AI control plane. This governance layer would enable token usage observability, track AI unit economics to connect spending directly to business value, and establish governance for models and automated code.
This proposal comes as the financial scale of the industry expands rapidly. Morgan Stanley estimates that generative AI revenue could grow from $45 billion in 2024 to more than $1 trillion by 2028. Dasgupta intends for this framework to shift organizational focus away from simple token expenditure toward operational discipline and measurable business outcomes.