Ramp Hits $44B Valuation as AI Spending Spirals Out of Control

Corporate AI spending becomes a budget crisis

Corporate America faces a new financial challenge: artificial intelligence costs are ballooning faster than finance teams can track them. Ramp, a spend-management platform, announced a $750 million funding round at a $44 billion valuation on Thursday, capitalizing on this exact problem.

The round was led by ICONIQ, GIC, and Ontario Teachers' Pension Plan, marking a 38% increase in the New York-based company's valuation. Ramp has crossed $1 billion in annualized revenue with positive free cash flow, according to CEO Eric Glyman.

Why it matters

AI token spending represents an entirely new category of corporate expense that most CFOs didn't anticipate in their annual planning. Unlike traditional software subscriptions with predictable costs, token-based AI usage scales unpredictably with employee behavior. Companies that fail to optimize this spending risk overpaying by orders of magnitude while competitors gain efficiency advantages.

The token spending problem

Glyman told CNBC that tokens—the units AI companies use to measure usage—cost considerably more than most finance leaders expected. The issue isn't just the absolute cost, but the lack of visibility and control.

"Suddenly you have this third pillar that has showed up, which is spending through tokens and intelligence," Glyman said. "It's not a clean area of spend."

Most companies default to using frontier models like those from OpenAI and Anthropic for every task, regardless of complexity. That's like hiring a surgeon to apply a bandage. Ramp's new product helps route tasks to appropriate models, potentially cutting costs by 99% for simpler operations.

The ROI question

The spending does correlate with results—when done efficiently. Among Ramp's 70,000 business clients, those spending the highest percentage of revenue on AI saw 12% revenue growth. Companies spending the least on AI experienced flat growth.

But efficiency matters enormously. Glyman noted that frontier model providers have no incentive to recommend cheaper alternatives. "Your incentive is truly to maximize revenue and profit," he said.

The end of tokenmaxxing

Some organizations adopted "tokenmaxxing"—using as many tokens as possible as a productivity proxy. Glyman believes this approach is dying as companies recognize that token volume doesn't equal business value.

"I think it's the twilight moment of tokenmaxxing," he said, adding that companies are becoming more sophisticated in their evaluation metrics.

For now, AI spending isn't replacing traditional software budgets, though Glyman expects that shift eventually. "I do think the bill will come due," he said.

The details were first reported by CNBC.