Corporate finance just changed forever, and it has nothing to do with standard software subscriptions. For the last five centuries, managing company cash boiled down to tracking two things: the people you employ and the vendors you pay. Suddenly, a third pillar has arrived. It is machine intelligence, it is billed by the token, and it is completely wrecking corporate budgets.
The chaos under the hood explains why spend management platform Ramp just secured a massive $750 million Series F funding round, skyrocketing its valuation to $44 billion. Led by major heavyweights like ICONIQ, GIC, and the Ontario Teachers’ Pension Plan, this raise marks a nearly six-fold valuation jump from just over two years ago.
This is not a story about a corporate credit card company getting lucky during a tech bubble. It is a direct response to a massive enterprise crisis: companies are burning millions on invisible AI infrastructure, and they have no idea how to track it.
The Out of Control Cost of AI Tokens
Most corporate software follows a predictable SaaS model. You buy fifty seats of a CRM, you pay a fixed monthly price, and your accounting team schedules the invoice. AI does not work that way. When your team builds automated workflows or connects internal data to large language models, you pay per token.
A token is essentially a fragment of a word. When your engineering team sets up an AI coding assistant or a customer service bot, every single prompt, response, and background data process consumes tokens. It is a pay-as-you-go utility bill, much like electricity, but vastly more volatile.
The reality on the ground is wild. Uber burned through its entire 2026 AI coding budget in a mere four months. Walmart had to place hard caps on its internal AI assistant to keep costs from spiraling. According to actual transactional data from Ramp, the median company utilizing AI now dedicates nearly 15% of its entire software budget strictly to AI tools. Monthly AI spending across enterprises quadrupled between early 2025 and 2026.
AI Spend Growth (Feb 2025 - Feb 2026): 4x Increase
Median Software Budget Dedicated to AI: 15%
The problem is that traditional accounting software is totally blind to this. If an engineer links an app to an API, the usage spikes silently. One in four major AI spenders experiences a cost surge of 50% or more in a single month. Finance teams do not find out until the invoice hits weeks later, creating a massive, multi-billion-dollar blind spot.
Why Legacy Finance Systems Are Breaking
If you have ever managed an enterprise budget, you know that procurement departments love predictability. They want contract reviews, set terms, and clear purchase orders. Token-based AI architecture breaks every single rule of traditional procurement.
- Fragmentation: A year ago, companies just bought OpenAI access. Now, they are testing multiple models simultaneously. Businesses are shifting daily between options like Anthropic, xAI, Together AI, and Lovable depending on project needs.
- Invisible Scaling: A single loop in an unoptimized autonomous AI agent can make millions of API calls in a weekend. An employee does not need a corporate card to trigger this; they just need access to the development environment.
- Delayed Invoicing: Because the majority of AI infrastructure is billed via invoice after the fact, finance leaders are constantly looking backward. You cannot manage real-time costs with month-old data.
This operational headache is exactly what is fueling Ramp’s hyper-growth. The platform surpassed $1 billion in annualized revenue and is generating positive free cash flow, processing over $100 billion in transactions annually. Its total payment volume jumped roughly 170% year-over-year as of March 2026.
Ramp is positioning itself as the literal financial infrastructure for the AI economy. They are launching dedicated token spend management tools that ingest billing data directly from AI providers. For the first time, a CFO can look at a dashboard and see tokens translated into actual dollars mapped to specific teams, projects, and use cases.
Automating the Back Office to Survival Mode
The funding round also highlights a massive shift in how finance departments operate internally. Ramp is deploying its own agentic AI tools to handle heavy administrative lifting. They recently launched Ramp Stack, an AI operating system aimed directly at accounting firms to handle account reconciliation, journal entries, and financial statement compilation.
The strategy is brilliant. By selling directly to accounting practices, they gain an immediate distribution channel to thousands of mid-market businesses.
Within corporate walls, the efficiency gains are becoming undeniable. In May 2026, the median company using these types of automated systems saved 50% more cash and 32% more hours compared to the previous year. For businesses utilizing integrated procurement and accounting agents, those savings more than doubled.
Ramp’s internal operations show just how aggressive this shift is. They use an internal software factory called Inspect that autonomously writes over two-thirds of the company’s own production code. Their internal AI workspace adoption sits at 99.5%.
Some market analysts argue that a $44 billion valuation is stretched, pointing out that competitor Brex was acquired by Capital One earlier this year for a fraction of that scale. But evaluating this market through the lens of traditional corporate cards misses the point entirely. The traditional spreadsheet-driven finance stack is undergoing its biggest structural shift in decades. The companies winning this space are software infrastructure plays, not banks.
How to Rein in Your Company's AI Spend Right Now
You cannot afford to wait for your next quarterly review to fix your AI leaks. If your teams are actively building with or using machine intelligence, you need to implement hard controls immediately.
Audit Active Vendors and Shadow IT
Do not punish your employees, but issue an immediate call for transparency. Scour expense reports, employee reimbursements, and accounts payable histories to build a definitive registry of every AI model, wrapper, and developer tool currently running. You will likely find three different teams paying for separate, redundant subscriptions.
Enforce Strict Virtual Card Guardrails
Stop letting employees attach general corporate cards or personal accounts to AI billing platforms. Issue dedicated virtual cards for specific AI vendors. Set rigid monthly spend caps directly on the card level. If a development team hits their cap due to an unoptimized model loop, the card declines, preventing a catastrophic five-figure surprise bill.
Implement Tiered Approvals for Variable Contracts
A $30 a month subscription does not need executive oversight, but open-ended API access does. Create a separate procurement track specifically for token-based infrastructure. Set automated approvals for minor software seats, but mandate strict, tiered approval thresholds the moment an invoice relies on usage-based metrics.
The corporate financial playbook is being rewritten in real time. If you keep trying to manage API tokens and autonomous agents with legacy accounting tools, your budget is going to get swallowed whole. Get your visibility sorted today before the next invoice arrives.
