Mavvrik Finds 25% Of Firms Cancel AI Projects Over Surprise Token Costs

The Common Question

Why are companies canceling AI projects and how common is it?

The Lesson in This Case

One in four companies has already delayed or canceled an AI project because unexpected costs forced the decision – not because the technology failed, but because no one could see the spending coming. Token-based pricing, shadow IT adoption by engineering teams, and consumption-based licensing have combined to make AI budgets nearly impossible to forecast, with only 11% of organizations able to predict AI spending within a 10% margin. Senior leaders who have not yet built unit-economics discipline into their AI programs – linking token consumption directly to measurable business outcomes – are running the same financial risk that caught cloud budgets off guard a decade ago.

Originally reported by CFO Dive. Read the full story here.