Firm scraps $60,000 marketing hires for one AI-augmented staffer

The commentary's argument runs against the grain of how companies have reacted to their first large AI bills. Accenture's July guidance warned against blanket usage cuts and noted that in nearly every deployment it examined, fewer than 10% of users and workflows drive the vast majority of the bill. By October 1, according to an internal announcement obtained by 404 Media, Accenture began centrally funding $29 a month per person toward Claude, with anything above charged to a project — a step Accenture says reporting that it imposed a firm limit is incorrect. The authors point out that fixed-price contracts are more than 65% of Accenture's bookings, so token costs land on its own margin, and that its own paper records hundreds of daily requests to lift individual spending limits.

The authors' alternative is to price finished work rather than consumption. They cite Jellyfish's study of 12,000 developers at 200 companies, where the median developer ran about $52 a month in tokens and the 90th percentile $691, and note the study is observational, showing association rather than cause. They also cite their own modeling of deployments in four industries, where the token bill was about a quarter of total cost. Their warning about premature rules draws on New York's 2015 licensing requirements for cryptocurrency firms, which they say drove a wave of exchanges away for good.