Exclusive: OpenAI's CFO pitches a new way to measure AI's value
Add Axios as your preferred source to
see more of our stories on Google.
Illustration: Sarah Grillo/Axios
As executives question whether their massive AI bills are paying off, OpenAI CFO Sarah Friar is pitching a new scorecard for measuring AI's success, according to a new blog post viewed by Axios.
Why it matters: The framework is OpenAI's response to a growing enterprise AI cost reckoning, as companies increasingly route work to cheaper models and demand clearer returns on their investments.
Zoom in: Friar is proposing a new enterprise AI metric: "useful intelligence per dollar." Instead of measuring AI spend by expenses or cost per token, she argues businesses should ask four questions:
- Is AI completing work that matters? Think customer issues resolved, code shipped or contracts reviewed, not just lots of usage.
- What does each successful task cost? Measure the full cost of a task, yes including AI usage and retries but also any cost of human review, rather than token prices alone.
- How often does it get the work right? The fewer corrections or escalations to humans, the greater the potential return.
- Does each AI dollar produce more value as usage grows? Companies should track whether AI is completing more high-quality work over time without exponential cost growth.
Between the lines: OpenAI is trying to shift the conversation from the price of AI to the value it creates.
- If companies judge AI by the amount of useful work completed rather than benchmark scores or token costs, frontier AI companies can make the case that more capable — and often more expensive — models ultimately deliver better economics.
Zoom out: CFOs are widely flying blind when it comes to their AI costs.
- One executive oversaw a half a billion dollar accidental Claude bill over the course of one month, as Axios reported.
- OpenAI CEO Sam Altman said costs are the second biggest challenge customers talk to him about, behind AI deployment within their organizations.
Reality check: OpenAI has an incentive to steer customers toward measuring outcomes instead of sticker prices, since its models are among the most expensive.
- OpenAI argues its most capable models will prove cheapest given the efficiencies associated with using top-tier intelligence.
- "The basic economic question facing CFOs and other business leaders is whether the value of the work AI completes grows faster than the cost of producing it," Friar writes.
- She argues businesses shouldn't just flock to the cheapest AI, but they should buy the AI tools that produce the most value, delivering on both costs and performance.
Yes, but: Many enterprises have already reached this conclusion, but they're responding differently than OpenAI might like.
- Rather than defaulting to the most capable frontier model, executives tell Axios they're increasingly routing everyday tasks to the cheapest model that can do the job, using only frontier AI models for the most intensive tasks.
- Others are experimenting with open-weight models or AI routers that automatically choose the best balance of cost and performance for each task.
The bottom line: As businesses worry about AI spend, OpenAI hopes to change how budgets for this technology are calculated.
