Amazon spent two years telling its people to use more AI. Then its own staff ran with that idea, and it backfired.
Some workers started burning through AI just to look busy. That ran up the bill.
So Amazon pulled the plug on the scoreboard that egged them on.
Amazon made an inside ranking tool called KiroRank. It ranked staff by how many AI tokens they used.
Tokens are the small chunks of words an AI reads and writes, and each one costs money to run. The board was meant to cheer on people who leaned into the tech. Instead, staff found a loophole.
They could climb it by pushing busywork through the AI, which ran up the bill while solving nothing.
Workers gave the habit a name: "tokenmaxxing." The Financial Times broke the story, and Amazon has since confirmed the board is gone.
For anyone tracking where the AI spending really goes, Market Briefs breaks it down each morning, plus a free investing class when you join.
Picture AI tokens like a taxi meter. It never stops running, so the more you use, the bigger the fare.
A board that rewards heavy use just rewards a bigger tab.
Dave Treadwell, a top boss at Amazon, told staff to cut it out. "Please don't use AI just for the sake of using AI," he said.
He wants staff to use it to solve real problems, not to chase a score.
Amazon says KiroRank was a side project built by a small group, not an official tool. It has now been shut off.
The firm still tracks token use to watch its costs. It just doesn't want staff gaming the number.
Token use has shot up across tech this year. A big reason is agentic AI, where a bot runs on its own for hours with little human help.
Each of those hours burns tokens, so the bills add up fast.
Amazon was already watching this in April. Its retail arm tracked how many engineers used AI each month and what came out of it.
The KiroRank move fits the same theme: less hype, more proof.
This isn't only an Amazon story. Meta took down a similar board called "Claudeonomics." Uber bosses have also said their AI spending isn't paying off the way they hoped.
One Uber leader even blew through the firm's AI budget for the year by April.
The big shift is in the question itself. For two years, the message was simple: use more.
Now the same firms, even the ones that make the chips that run AI, are asking if any of it works. Amazon is one of the first to say enough.
For investors, that mood swing is the part to watch. Heavy AI spending was a badge of honor not long ago.
Now it's a cost that has to earn its keep, and that change shapes how these firms guide on profits.
Want this kind of read every weekday? Join the hundreds of thousands reading Market Briefs and get a free 45-minute investing course too.
