Ramp Targets Token Spend Management
$1.5B/year corporate card neolab
Ramp is trying to own the finance team’s new fastest growing line item before it becomes a standalone category. Cards got Ramp into every transaction, bill pay and procurement let it see invoices and contracts, and token spend management extends that same control layer to AI usage, where finance teams need to see which model, team, and workflow is driving costs before token bills sprawl across the P&L.
-
Ramp’s expansion has followed a clear path from card swipe to full spend graph. It started with interchange on card volume, then added bill pay, procurement, travel, and treasury, which widened revenue beyond interchange into subscriptions, payments fees, affiliate fees, and deposit economics. Token spend fits this pattern because it is another spend stream that finance wants governed inside one system.
-
The product logic is the same as Ramp’s earlier AI push. Ramp uses AI to turn messy finance inputs like receipts, invoices, and contracts into structured fields, then automate approvals, policy checks, renewals, and vendor analysis. Token spend management applies that workflow to OpenAI, Anthropic, and gateway usage data, so finance can break down cost by employee, project, model, and use case instead of staring at one large monthly API bill.
-
This also sharpens Ramp’s positioning versus adjacent rivals. Mercury is converging on product from banking upward, while Ramp stays on the application layer and keeps adding software and AI control surfaces on top of partner infrastructure. That makes token governance a natural move for Ramp, because it looks more like finance software than like a bank product.
The next step is for AI spend to become managed like payroll, travel, or SaaS, with budgets, approvals, anomaly detection, and optimization built in from day one. If inference keeps moving from experiment to operating expense, Ramp can become the default system where companies decide not just how money is spent, but which machine labor is worth buying.