
UpTrajectory Review
Fast Company flags a quiet budget crisis unfolding inside companies that rushed to hand employees AI tools: most firms cannot say what AI actually costs them. Citing an unreleased KPMG survey, the piece reports that only one in four companies has a comprehensive view of AI spending, half have partial visibility, and one in five discover the damage only when the invoice arrives. KPMG's global AI lead Steve Chase frames AI as a new resource that needs active management, and his firm now works with companies that burned through annual token and cloud budgets in months.
For a small-business operator, this story reads like a warning shot. Large enterprises absorb a surprise six-figure bill; a ten-person shop cannot. The piece describes employees "tokenmaxxing"—using AI flat out—while the person paying sits far from the person spending. That separation of user and buyer is the core problem. If your team treats AI credits like free office coffee, you will find the true cost in your monthly statement, not in your budget spreadsheet.
The genuinely new angle here is the unit of cost itself. AI bills run on tokens, fragments of text or code that models process, and tokens defy normal budgeting. They do not map cleanly to words, some can be cached and skipped, others process fresh each time, and the total stays fuzzy until the provider tallies the month. This is not like metering kilowatt-hours or phone minutes. It is closer to paying for fog. That novelty explains why finance teams lag and why the KPMG numbers look so poor.
We agree with the piece's skepticism toward blind adoption, though it underplays one fix: usage limits and per-seat caps are standard controls most platforms already offer. The Axios example it cites—half a billion dollars in one month after a client left Claude licenses uncapped—sounds extreme, but it illustrates a preventable failure, not an inevitable one. The deeper issue is governance. Companies rolled out AI for fear of missing out, then discovered that freedom without guardrails produces waste. Shadow AI spending now mirrors the old problem of shadow IT, only the meter runs faster.
The second-order effects reach beyond finance. Employees who build workflows on unlimited AI may resent new limits, and heavy users often drive the most valuable automation. Cutting them off bluntly could stall real productivity gains. Procurement teams, meanwhile, face a new task: negotiating contracts for a resource whose unit price shifts with caching behavior and model updates. Smaller firms feel this sharply because they lack dedicated AI ops staff to tune prompts, route queries to cheaper models, or spot redundant usage.
Watch for the tools vendors ship next: dashboards that attribute token spend by team or project, and pricing models that move toward predictable subscriptions. Until then, audit your own usage this week. Pull last month's AI invoices, identify your heaviest users, and ask what they built with the spend. Set per-person limits, require pre-approval for large workloads, and assign one owner to review the bill monthly. Treat AI like any other utility: measure it, cap it, and make the user feel the cost.
“They turn on usage, and suddenly the people paying the bill are not the people using the product.” — Fast Company
Takeaway: Audit your AI bills now, set per-user spending caps, and assign one owner to review usage monthly before costs spiral.
Excerpt from the original — Fast Company
FOMO is strong around AI, with companies adopting the technology willy-nilly. Many have encouraged employees to tokenmaxx to their hearts’ content. Now they’re starting to realize that freedom comes with a price.
Just one in four companies say they have a comprehensive view of what artificial intelligence is costing them, according to an as-yet-unreleased KPMG survey reported by The Wall Street Journal. Only about half have even some visibility into the cost of their AI use. One in five have no visibility, or only see the damage once the bill arrives. “It’s a new resource that needs to be managed that didn’t exist quite that way, and we’re seeing exponential growth,” Steve Chase, KPMG’s global head of AI, told the Journal.
Part of the problem is pinning down what, exactly, AI costs. The basic unit of AI use—the token—is an unusual thing to budget around. Each token is a fragment …