A former CFO perspective on AI emphasizes financial control, forecasting, and ROI, noting that model inference costs have dropped roughly 75% in the last 18 months and are likely to continue falling. To manage costs, organizations should choose the right model, as 95% of enterprise AI still runs on the most expensive models, yet aligning model size to task complexity can cut inference costs by 60%–80%. Eliminating waste through caching can reduce redundant processing by 40%-60%. Treating prompts as code—versioned, tested, and optimized—can yield consistent cost reductions of 30%–50% without sacrificing output quality.