$500 Million in One Month: What Happens When AI Has No Guardrails

In May 2026, an unnamed company faced a significant financial setback by burning through $500 million in Claude credits within a month due to a lack of governance and spending limits, highlighting a broader issue of unregulated AI usage across industries. This incident, along with Uber exhausting its annual AI budget by April and Microsoft canceling Claude Code licenses, underscores a structural problem where companies have failed to implement proper governance layers to manage AI resources effectively. Without clear usage limits, visibility dashboards, and approval workflows, businesses are prone to excessive and inefficient AI consumption, reminiscent of early cloud adoption challenges. The situation calls for a platform engineering approach to AI governance, ensuring consumption is visible, accountable, and tied to outcomes rather than mere usage metrics. This shift from ungoverned to governed AI usage is necessary to prevent financial surprises and ensure that AI tools deliver their promised value efficiently.