Amazon Pulls Back on Big AI Spending

Some internal projects cost far more than expected when switched from traditional code to AI. In one case, a single task spent nearly two million dollars on AI models without ever finishing. Senior staff now call these results too expensive to justify.

The company is putting new limits on how teams use these tools. Engineers want to make sure AI actually saves time and money instead of wasting it. They plan to keep a closer eye on project budgets moving forward.

Leaders shared these lessons at a recent company meeting. They are encouraging staff to focus on cost rather than just using new tech for every problem. This shift signals a move toward more practical uses of AI across the business.