OpenAI released usage analytics and spend controls for ChatGPT Enterprise on June 18, responding to a string of enterprise AI cost overruns that have left CFOs scrambling for answers. An Axios investigation revealed one large client incurred $500 million in AI services charges in a single month after failing to set a cap. Uber exhausted its entire 2026 AI coding budget by April, its CTO told The Information, as engineer usage surged from 32% to 84% in three months.
Every prompt, response, and autonomous agent action consumes tokens, the basic billing unit for AI models. These costs accumulate without a deliberate spending decision, so usage can spike long before finance teams notice. Global AI spending is on pace to hit $2.59 trillion this year, a 47% jump, according to Gartner, and tokens represent a growing share of that tab.
A governance gap exposed
Uber's experience is not an outlier. Microsoft revoked its own developers' AI coding licenses months after granting them, and a Priceline employee told TechCrunch that a routine coding tool contract renewal came back four to five times more expensive than budgeted. In each case, actual usage outstripped internal tracking, and autonomous AI agents burned through tokens far faster than simple chat sessions.
What OpenAI's new dashboard delivers
The Global Admin Console now combines ChatGPT and Codex credit usage into a single view, breaking down consumption by user, product, and model. Companies can finally trace spending back to the workflows generating it, rather than discovering the total after the invoice arrives. Admins can set default credit limits for an entire workspace, configure separate limits for specific teams, and create individual overrides for employees who need more capacity. Employees can check their own usage against their budget and request more credits with context about the task, a structure that mirrors how organizations already manage cloud computing bills.
Industry push for token standards
On June 3, the Linux Foundation announced its intent to launch the Tokenomics Foundation, a nonprofit body built to standardize how the industry measures and governs AI token spend. The scale problem behind the effort dwarfs any single company's overrun. Global token usage is projected to multiply 24 times between 2026 and 2030, reaching 120 quadrillion tokens a month, while the inference market underneath it grows from roughly $106 billion in 2025 to $255 billion by 2030, according to Goldman Sachs research cited in the foundation's release. Backers include Accenture, Google Cloud, IBM, JPMorganChase, Microsoft, Oracle, Salesforce, SAP, and Booking.com.
Chris Reed, senior director of IT finance at Booking.com, said the group "needed neutral, vendor-independent standards instead of relying on usage data each AI provider chooses to disclose."
Why this matters for finance leaders
Ramp's billing data illustrates how uneven the cost exposure already is. The median business paid $2,246 a month for AI tokens in April 2026, but the average reached $140,842 - a gap that shows how one ungoverned workflow can push a company's bill far past what its budget assumed. Investors are reading the signal clearly: AI vendors and cost-management startups are now competing on governance as much as on model capability. For finance teams building internal AI governance, structured learning like the AI Learning Path for CFOs can help translate technical token metrics into budget controls that a CFO can approve.
Your membership also unlocks:
