Ramp Raises $1.85B at $60B Valuation to Build AI Finance Layer
Ramp has raised $1.85 billion at a $60 billion valuation as it pushes deeper into AI. The round was disclosed on Oct. 15, four months after the company raised $750 million at a $44 billion valuation. TechCrunch
The financing was co-led by Dragoneer Investment Group, Thrive Capital, Founders Fund, Sands Capital and Avenir Growth. Ramp said total funding now stands at about $5 billion. Capital is not the constraint. Execution is.
The numbers behind the step-up
Ramp said annualized revenue has crossed $2 billion, up from $1.5 billion in June 2026. Growth is 25% faster than around the same time last year. The company said it now facilitates over $275 billion in annualized purchase volume, up 175% from a year earlier. Customer count stands at 85,000, up from 70,000 in June. TechCrunch
Customers include Visa, Uber, Shopify, Robinhood, Anduril, Notion, Cursor and Figma. The mix spans payments infrastructure, logistics, commerce, brokerage, defense technology and productivity software. For a spend platform, that breadth matters. It feeds transaction data across categories, procurement cycles and company sizes.
The cadence is unusual even by late-stage standards. Four months separate a $44 billion round and a $60 billion round. Revenue moved from $1.5 billion to $2 billion annualized in the same window. Customer count added 15,000. Purchase volume kept climbing while growth accelerated.
From expense management to agent infrastructure
Ramp has expanded beyond expense management into payments, fraud detection, procurement, vendor management and accounting. The expansion reframes the product as a finance operations stack rather than a card with software attached.
The company has embedded AI agents into procurement, expense management, accounting and budgeting products. The agents operate inside approval, coding and reconciliation workflows where policy and spend controls already live. That placement is deliberate in architectural terms. Agents do not sit alongside finance software. They sit inside it.
Ramp also launched a corporate credit card designed for AI agents to use, giving programmatic actors a credentialed payment method under existing corporate controls. In August 2026, the company entered the model routing business with an AI model router called Router. It then launched Glass, a workspace that collects and remembers AI automations, prompts and skills built by an organization's employees. TechCrunch Taken together, the releases cover transaction execution, model selection and organizational memory. One handles how agents pay. One handles which model answers. One handles what the organization retains.
Ramp described the milestone in a blog post titled "Ramp at $60B: The End of the Beginning" at https://ramp.com/blog/ramp-at-60-billion. Ramp Bloomberg had reported in September 2026 that the company was in early talks for a new round at roughly a $60 billion valuation. Bloomberg
The broader context here is the shift from tracking spend to authorizing it in software. Expense tools started as systems of record. Procurement added workflow. Cards added a payment rail. Once agents can initiate purchases, draft vendors, code invoices and reconcile ledgers, the control point moves. It is no longer the monthly close. It is the moment of delegation, when a human or another agent grants authority to act and spend.
Looking at what this means for technology teams, the interesting questions are identity, policy and observability. An agent with a card looks like a new kind of principal, one that needs scoped credentials, spending limits, merchant controls and a complete audit trail. Model routing looks like a parallel governance problem, choosing models against latency, cost and quality constraints per task rather than per vendor contract. Glass points to a third problem, how prompt libraries, skills and automations persist when employees change roles or leave. Finance data may prove a useful substrate for all three, because every agent action that costs money leaves a structured trace.
In this author's view, the valuation math reflects that platform logic more than card economics. Purchase volume at $275 billion annualized and revenue at $2 billion annualized imply monetization on flow plus software subscription, with growth still accelerating. The jump from 70,000 to 85,000 customers in four months suggests mid-market and enterprise adoption is broadening, not just deepening inside existing accounts. Worth flagging, that same breadth raises the bar for reliability, access controls and reconciliation accuracy. A procurement agent that misclassifies spend is an annoyance. One that commits spend incorrectly is an incident.
What this enables, if the execution holds, is a finance stack where policy enforcement, payment and ledger update happen in the same loop. That has been the long promise of autonomous back-office software. The pieces here, credentialed agent payments, embedded workflow agents, routing across models and a shared memory layer for automations, map closely to what practitioners say has been missing. The work ahead is less about model capability and more about trust boundaries, exception handling and clean integration into ERP and data warehouse realities.
