Businesses are confronting steep AI bills as the technology transitions from experimental tool to expensive necessity. Companies once encouraged unrestricted AI experimentation, but major providers have shifted to usage-based pricing models to demonstrate revenue sustainability ahead of potential IPOs. The change has caught many organizations off guard.

Uber revealed it exhausted its entire 2026 AI budget by April, while a KPMG survey of 2,145 business leaders found nearly half scaled back AI agent deployment due to costs exceeding benefits. OpenAI's Sam Altman confirmed this represents a rapid shift from non-issue to major concern. Companies are now implementing strict controls on AI spending.

Atlassian caps employee token usage and requires manager approval for increases, while Hostinger benchmarks models to optimize costs across its 900 employees. Vendor lock-in concerns are prompting businesses to adopt routing tools that select the most cost-effective models for specific tasks rather than relying on single providers.

Open-source alternatives offer potential relief, with Deloitte estimating they can deliver 90% of closed-model performance while reducing costs by up to 70%. Despite tighter controls, Goldman Sachs projects a 24-fold increase in global token consumption by 2030 as AI agents proliferate. Organizations must balance innovation with fiscal discipline to avoid budgetary blowouts.