According to a report from the Wall Street Journal, corporate leaders have realized that burning AI tokens actually costs money, and doing it almost exclusively with no other goals in mind isn’t really a great business strategy. The good thing is that these people get paid millions of dollars a year to figure these things out.
Earlier this week, Uber CEO Dara Khosrowshahi said that the cost of AI initiatives within the company were “becoming hard to justify” because the output was not commensurate with the token burn rate, while acknowledging the reason they were doing so much work on token burning in the first place is that it “can feel” as if AI is free.
Not so, as it turns out.
An anonymous AI consultant told Axios that one of its clients accidentally spent half a billion dollars in a single month because it never bothered to put usage limits on employee access to Anthropic’s cloud. That means…a lot. To the point of straining credibility. The Journal didn’t find anything that serious, but did hear about one financial institution that saw its employees spend hundreds of thousands of dollars worth of tokens per month using premium-tier models to ask employees basic questions and lead to absurd back-and-forth conversations.
This very foolish era of justifying AI expenditure was always going to go this way. Corporations have already spent a lot of money adopting these systems, and they need to justify this expense. To do this, they encourage their employees to use it as much as possible. In turn, employees do the same—even when it doesn’t make sense to use AI for a task.
Meta scrapped its token-burning leaderboard last month after it was leaked, revealing that top “token legends” managed to burn 281 billion tokens in a month – 33 times more than the amount of computation needed to reproduce the entirety of Wikipedia. According to the Financial Times, Amazon joined that rollback this week, removing its scoreboard of employees who used the company’s internal AI tools the most — a decision that was reportedly made after it became clear that employees were giving meaningless tasks to AI agents just to keep their spot on the leaderboard.
It is clear that the corporate world is ready to burn money in an attempt to justify its current cash crunch. However, it turns out they have limitations. Before shareholders start wondering what the price tag of all those tokens is rising, you can just use “tokens burned” as a metric on many quarterly earnings calls.
<a href