One company spent $500 million in one month after forgetting to set AI usage limits

It has been an unexpected shift in views on AI, with companies recently rejecting its use due to unsustained production despite rising API costs. Recently, brand leaders such as Costco, Delta Airlines and IBM have echoed his concerns about AI and his preference for retaining the human workforce, especially as others, such as Amazon, Meta and Microsoft, continue to cut jobs. More recently, comments from Uber’s new chief operating officer, Andrew Macdonald, about AI-related costs and the use of tokens that does not improve worker productivity as it should was heard, and mostly appreciated, on the Internet. This was followed by information that Uber engineers had already exhausted their 2026 AI budget.

It turns out that Uber may not be the only company struggling to keep its AI budget in check. According to a axios report (paywalled), an unspecified company spent approximately $500 million on claudio credits after failing to put up barriers to use. This, among other incidents, is beginning to push corporate leaders to evaluate whether AI is truly delivering the value they initially assumed.

The report also notes that the corporate cadre is beginning to ditch “tokenmaxxing,” a term used to describe the tendency to consume AI credits as quickly as possible. To counter that sentiment, the AI ​​biggies, including Google, have been creating models and inference techniques that are more cost-effective.

Added to this, a recent Gartner Report says that inference costs for generative AI models in 2030 will be only a tenth of what they were in 2025. However, it is important to note that usage can also grow exponentially, especially as our reliance on AI agents increases and processes become more complex. The report also predicts that token usage will expand by 5 to 30 times current usage.

Vendors including Google and Anthropic have also recently moved to a usage-based billing and stricter use limitswhich has rightly caused agitation among non-business users.

Even companies betting their future on AI, like Microsoft, are moving away from their tokenmaxxing approach. Earlier this month, Microsoft reportedly began canceling Claude subscriptions and discouraging employees from using it too much, just six months after he began pressuring various workers of different profiles to code more in Vibe. This is further evidence that overreliance on AI is costing businesses more than it benefits.

It is difficult to predict whether we will be able to see a complete reversal of the initial fervor for AI. To be fair, it’s also highly unlikely. But we could certainly see companies budgeting their use of AI and restricting it to certain activities rather than giving free rein to employees. The AI ​​bubble, so to speak, may not burst, but the AI ​​dream could already be starting to end.