Several companies have faced significant financial setbacks due to AI token billing errors. A major incident involved a company spending $500 million on Claude in one month after failing to set usage limits for employees. Similarly, an Australian AI consultant incurred an $18,000 Google Cloud bill overnight due to an exposed API key, despite having a $7 budget. Additionally, Peter Steinberger's team at OpenClaw generated a $1.3 million OpenAI API bill over 30 days, though this was part of an internal experiment. The shift to usage-based billing models by companies like OpenAI and GitHub has led to increased costs for heavy AI users. GitHub's transition to usage-based billing, effective June 1, 2026, has sparked backlash from users facing steep price hikes. Meanwhile, some companies are developing solutions to optimize AI usage and reduce costs, such as Glean's enterprise AI assistants and Factory AI's model routing technology. As AI costs begin to rival labor expenses, companies are reevaluating their AI strategies to maximize value and efficiency.