Finance teams that pay OpenAI, Anthropic and Google directly for AI usage typically cannot answer a basic question: which team, project or API key is driving the bill. Ramp is targeting that blind spot with an expanded version of its AI Token Spend Management tool, announced Thursday, which now pulls usage costs from those three providers plus Cursor and OpenRouter into a single dashboard. The move treats AI usage as its own finance category rather than a line item buried inside a cloud bill.
Corporate finance has historically split spending into two buckets: payroll and vendor contracts. Ramp argues AI tokens form a third bucket that behaves unlike either. Usage-based pricing means costs move with consumption, not headcount. Charges attach to API keys rather than named employees, and monthly invoices from providers rarely break down what was actually used.
The expanded product connects to OpenAI, Anthropic, Gemini, Cursor and OpenRouter, letting finance teams slice spending by provider, by team, by individual, by project, or by API key. Administrators can cap spend at the team, project or key level and route overrun alerts to the people responsible before costs run past plan. A weekly briefing summarizes spending trends, and the platform reconciles vendor invoices against logged usage.
Thousands of businesses have already connected an AI provider account to Ramp, the company said. The company also disclosed that a third of customers have access to a cheaper model capable of doing the same job, a mismatch the dashboard is designed to surface so finance can push engineering toward the lower-cost option.
Ramp Chief Financial Officer Will Petrie cast the tool as more than a cost-cutting measure. “Managing AI spend is not just about controlling costs. It is about knowing where to invest next,” Petrie said. “Finance teams need to see which use cases are creating value, where spend is drifting, and where another dollar can drive the most growth.”
The clearest evidence of the tool’s value, per Ramp, involves AngelList Advisors LLC. A weekly briefing surfaced prompt caching, a technique that cuts redundant model calls, after the finance team had quietly bled $10,000 a month without anyone catching it. One of AngelList’s controllers said the tactic had not been on his radar. Engineering shipped a fix the same day the alert arrived.
Ramp built the underlying system first to manage its own finance operation’s AI spend before turning it into a customer product. Internally, the company says 99.5% of its employees now use AI tools every working day, and token costs have become a real operating expense rather than a rounding error. The dashboard costs nothing to start, whether or not the business already uses Ramp for expense management or corporate cards.
The expansion lands weeks after Ramp closed a large funding round. The company, founded in 2019 and headquartered in New York, secured the financing in June: a $750 million Series F that valued Ramp at $44 billion. Ontario Teachers’ Pension Plan, GIC and ICONIQ Capital backed the round, which pushed cumulative funding above $3 billion.
The broader shift here is category formation. FinOps emerged as a discipline once cloud compute bills got large enough that engineering alone could not be trusted to manage them; AI token spend is now following the same trajectory, moving from a line item inside a cloud invoice to a governed budget category with its own dashboards and alerts. For CFOs and CTOs, the near-term decision is not which vendor dashboard to adopt. It is who owns accountability for AI spend, before the monthly bill gets large enough that the org chart decides the answer by default.
SiliconANGLE (Duncan Riley) reported this story on July 16, 2026.
