FinOps is being rapidly rewired for the AI era, where token economics introduce unpredictability that cloud billing never had. Token prices are falling but enterprise AI costs keep rising because newer reasoning models consume more tokens per task. Finout CEO Roi Ravhon and Google Cloud's Pathik Sharma discuss how organizations should route requests to the cheapest capable model rather than defaulting to frontier models, and why agentic FinOps tools must use deterministic guardrails for critical actions like server termination. They also argue that FinOps is fundamentally an organizational culture problem — cross-team accountability and engineering cost awareness must come before any tooling investment.
Table of contents
Token economics, on a clockDon’t ask the LLM to fix your KubernetesIt was never about the tool112 Impressions