OpenAI CFO Proposes AI Investment Return Scorecard

In a recent article, OpenAI CFO Sarah Friar presents a scorecard for measuring AI investment returns. The scorecard includes four key metrics: useful work, cost per successful task, dependability, and return on compute. This framework helps organizations systematically assess the actual value of AI tools, models, or infrastructure. For enterprise decision-makers and AI project leads, the scorecard provides a clear method to quantify AI investment outcomes, avoiding reliance on intuition or vague metrics. For instance, tracking cost per successful task enables teams to compare efficiency across models or APIs, while return on compute helps optimize infrastructure spending. This is especially useful for teams procuring AI solutions or deploying models internally, aiding smarter budgeting and strategic choices.