started · updated
Corporate AI adoption faces rising token costs and governance challenges
As enterprises rapidly integrate generative AI, companies are encountering significant hurdles regarding token consumption costs and management governance.
Credit Saison reported a "token exhaustion" incident in June 2026 after distributing ChatGPT Enterprise to approximately 3,600 employees. The surge in usage caused the company to deplete its monthly credit limit by mid-month, forcing a shift to more restrictive, lower-cost modes to maintain service. To manage this, the company implemented a strategy similar to managing color versus black-and-white printing, encouraging employees to use high-performance features only when necessary.
Akamai Technologies highlighted broader global trends, noting that while model unit prices have dropped, total usage costs continue to climb. For example, Uber Technologies reportedly exhausted its annual AI budget in just four months, leading to individual monthly spending caps. The shift from human-led chat to AI agents has further complicated oversight, creating difficulties in tracking who is using which tools, what data is being sent, and how much is being spent. Akamai suggests that managing these risks requires integrated platforms capable of allocating, reducing, measuring, and controlling AI requests.
Entities
Akamai Technologies · ChatGPT Enterprise · Credit Saison · Kazutoshi Ono · Uber Technologies