AI is transforming enterprise operations. It is also transforming enterprise budgets — and most organisations are not ready. Token-based billing, GPU infrastructure costs, multi-model routing decisions, and the runaway economics of autonomous AI agents have created a cost management challenge that traditional cloud FinOps was never designed to address. AI FinOps: The Practitioner's Guide to Managing, Optimizing, and Governing AI Costs at Scale closes that gap. Written by a 30-year technology and cybersecurity presales veteran, this book is the definitive practitioner's handbook for everyone responsible for making AI investment financially accountable — FinOps engineers, cloud architects, AI product managers, technology finance leaders, and the CTOs and CFOs who need to govern AI spend at scale. Across twelve comprehensive chapters, the book delivers: Token economics and LLM pricing — understand exactly how API billing works, why context windows compound costs dramatically, and how prompt engineering decisions translate directly into monthly invoice line items. AI infrastructure economics — GPU selection, training cost estimation formulas, inference serving architectures, and the rigorous break-even analysis for the API vs self-hosting decision. Multi-model strategy — the model selection matrix, four routing strategies from rule-based to classifier-based, cascading architecture design, and fine-tuning economics that deliver 70–85% cost reductions with no quality loss. AI cost visibility — a twelve-dimension tagging taxonomy, AI gateway architecture options, the FOCUS specification for cross-provider billing normalisation, and a five-level visibility maturity model. The optimisation playbook — eighteen concrete levers with documented saving ranges, a prompt compression before/after worked example, caching strategy decision tables, and three real-world case studies delivering 59–96% cost reductions. Governance and guardrails — a ten-domain policy framework, federated organisational model, budget governance tiers, eight guardrail types, and a five-cadence operating rhythm. Agentic AI cost governance — the cost multiplication mechanics, P95 scenario cost modelling, eight agentic guardrail types, seven architecture pattern cost profiles, and a five-level agentic maturity model. AI unit economics and ROI — ten cost-per-outcome metrics, eight ROI value categories, industry benchmarks by use case, and executive communication frameworks tailored to CFO, CTO, board, and engineering audiences. Maturity model and roadmap — a comprehensive five-dimension, five-level maturity framework, a 12-question self-assessment scorecard, and a 90-day implementation roadmap with phase-by-phase KPIs. Grounded in the FinOps Foundation's State of FinOps 2026 data — confirming that AI cost management is now the most in-demand FinOps skill globally — this book equips practitioners to move from reactive cost discovery to proactive value governance. AI costs are not going to manage themselves. This book shows you how.