As AI adoption crosses 50%, the tokenmaxxing economy splits off and up

We're halfway there… living on a prayer Nvidia GPU. Business AI adoption just crossed 50% for the first time, according to the Ramp AI Index. March marked the midpoint of the vibe coding era: 50.4% of businesses on Ramp now pay for AI services, up from 35% a year ago. Paid adoption is a key indicator for AI’s future, since models and the data centers they run on are hugely expensive.

• Model behavior: Anthropic adoption surged from 24.4% to 30.6%, its largest month-over-month spike since Ramp started tracking this data three years ago. Anthropic has been swiftly narrowing the gap with OpenAI (35.2% adoption).
• The VC effect: Venture capital-backed businesses lead with 80% adoption as they influence portfolio companies to use the latest tools.

Move fast and spend tokens…The experimental phase of AI has given way to an urgency-tinged stage. Ramp data shows businesses' monthly AI spend quadrupled from February 2025 to February 2026. And unlike predictable seat-based SaaS pricing, the AI billing model is based on usage (a lot harder to oversee and control). If employees’ Claude Code token usage spikes, finance may get hit with a surprising bill.

• Tokenmaxxed out: Corporate pressure to adopt AI has spawned "tokenmaxxing" — when employees use as many tokens as possible. Several tech companies are using internal leaderboards to track token usage.
• Ramp data shows AI costs spike 50% or more about one in every four months for the biggest spenders.

A Claude for concern? Not for AI labs. The tokenmaxxing ethos is a very positive indicator for LLM providers. Cue: Anthropic said last week that the number of its business customers spending over $1M on an annual basis has doubled since February (to 1,000). And OpenAI recently closed Silicon Valley’s largest-ever funding round ahead of its OpenAIPO. Anthropic and xAI-owner SpaceX are also gearing up for anticipated mega-debuts this year. The IPO cash will be critical as AI labs reportedly burn through billions. Diversified AI players like Google, Amazon, and Meta, which are already highly profitable, have more cash burn cushion.

The bottom line:

AI’s next chapter is about ROI. Getting to 50% adoption was about experimentation. The next phase is about validation. Companies will be looking to clearly measure AI’s effect on revenue, profit, and productivity. While it’s still early, Ramp data indicates that businesses that spent heavily on AI tools grew their revenue much faster than those that didn’t. Of course, correlation ≠ causation. But based on the level of investment, expectations for AI’s ROI are high. If the numbers don’t add up once visibility sets in, a corrective pullback may follow.

For more stories like this, subscribe to the Leading Indicators newsletter.