Software Engineering Daily · Tuesday, August 11, 2026
Amazon reported significant unplanned spending, described as "catastrophically expensive," due to AI agent loops that were not properly crashing and continued to incur costs. This issue is linked to incentives that reward token consumption rather than actual value, a practice termed "token maxing."
“Amazon basically said that they had a ton of unplanned spend. They called it catastrophically expensive.”
“Apparently they had about 860% budget overrun over five months. And this was basically caused by bad agent loops just didn't crash loudly enough and they've just kept being built.”
“Like you're rewarding people for token consumption. So it's like you're giving people a license to be wasteful. And not looking at the productivity metrics.”