OpenAI’s Chief Financial Officer, Sarah Friar, recently shared her framework for assessing the economic impact of investments in artificial intelligence. Unlike traditional software evaluations, which focus on metrics such as user counts and renewals, Friar emphasizes that AI effectiveness should be gauged by the actual value it generates.
In her blog post, Friar posed a crucial question for companies: Is the value derived from AI's outputs increasing at a pace that outstrips the costs incurred to generate them? She stresses that understanding this requires moving beyond surface-level statistics like cost per token.
Friar introduces a vital metric termed “useful intelligence per dollar.” This evaluation encompasses four components: the significance of the AI's output, the cost associated with each successful task, the reliability of the results, and whether increased usage enhances the value derived from each dollar spent.
To operationalize this, Friar advises that leaders should monitor the volume of quality-assured tasks completed by AI, total the expenses involved, and calculate the cost per successful task. They must also verify if stakeholders can rely on these outputs and whether the growth in high-quality work occurs at a rate that exceeds total costs, all while maintaining or improving quality. If these conditions are met, it indicates that each dollar invested in AI is yielding greater returns, with computational power playing a pivotal role in this dynamic.
“Our mission is to enhance this equation in every iteration: evolving better models, delivering quicker and more reliable outcomes, and reducing costs for essential tasks,” she articulated.
For OpenAI, compute resources are not merely an operational cost but a strategic asset. While the company does not publicly disclose detailed capital expenditure plans due to its private status, it had previously revealed the Stargate initiative, which outlines a comprehensive investment plan of $500 billion over four years to develop expansive AI infrastructure in the U.S. The initial funding phase aims for $100 billion, with a broader target of achieving a 10-gigawatt capacity by 2029. Remarkably, OpenAI has already surpassed its early milestones given the growing demand for AI solutions. Speculations suggest that the company could file for an IPO this summer or as late as 2027, with its current valuation standing at approximately $852 billion and poised to approach $1 trillion.
As Chief Financial Officers take on increasingly strategic roles, they are now expected to collaborate closely with CEOs in determining significant long-term investments, particularly in areas such as AI.
Recently, McKinsey hosted its 24th Global CFO Forum, attracting around 100 finance executives from over 30 countries, representing leading organizations. Andy West, a senior partner at McKinsey and co-leader of its Strategy and Corporate Finance division, shared insights from an informal survey he conducted during the event, finding that about two-thirds of CFOs now oversee strategic functions, an increase from less than a third five years ago.
“AI has been a topic of discussion at this conference for a few years,” West noted. He observed a shift from last year’s experimentation with AI toward an emphasis on transformation across entire enterprises in the current discussions.
