Claude is now the most-used paid AI tool in U.S. businesses, at least by one measure. April 2026 data from Ramp’s AI Index puts Anthropic’s Claude at 34.4% of business AI spend share, with OpenAI’s ChatGPT at 32.3%. Anthropic gained 3.8 percentage points month-over-month. OpenAI lost 2.9. If the trend holds in May, Claude’s lead will widen past statistical noise.
Before this gets read as a coronation: it isn’t one. Ramp runs a corporate spend management platform. Its AI Index tracks what companies pay for AI tools, not seat counts, not active sessions, not tasks completed. A single enterprise team renewing an Anthropic contract at a higher tier moves the index more than a thousand SMBs switching chatbots. The methodology is sound for what it measures. What it measures is spend share, not adoption breadth. Keep that distinction close.
That said, the direction is meaningful. Three months ago, OpenAI was the default assumption for enterprise AI procurement. Anthropic was the challenger with a compliance story and a model that developers liked. Something shifted. The Ramp data is the first spend-side signal that the shift has crossed into purchasing behavior, not just preference surveys.
The context matters here. Anthropic has made deliberate enterprise moves in the past 90 days: tighter API pricing for high-volume contracts, Claude Code’s rising adoption among development teams, and a series of institutional deployments that include defense-adjacent infrastructure. The company that recently closed what’s been reported as a $30 billion financing round isn’t just building models; it’s building distribution. Enterprise procurement teams are responding to that.
OpenAI isn’t standing still. Its Deployment Company structure, backed by a consortium of private equity investors and confirmed in prior reporting, is designed specifically to extend OpenAI’s reach into enterprise environments it can’t serve at full margin today. But deployment infrastructure takes time to produce revenue. The April spend data suggests that time window matters.
Overall AI adoption across the index increased 0.2 percentage points in April. The market isn’t expanding dramatically month over month. That means Claude’s gain is largely OpenAI’s loss, this is share competition, not a rising tide.
What to Watch
What to watch
the May Ramp AI Index is the first real test of whether April was a trend or a spike. A second consecutive month of Claude gains above 2 percentage points would signal structural shift. A reversal would suggest April was noise, a large contract renewal, a cohort of new enterprise Claude deployments timed to Anthropic’s financing close, or seasonal procurement patterns. The index won’t tell you which without the underlying transaction data. Ramp doesn’t publish that.
The real story is what enterprises do with this signal. Procurement teams that have been running OpenAI as their default may now have organizational justification to run a competitive evaluation. Enterprise AI is outperforming consumer AI on revenue metrics for reasons that map directly to this dynamic. That’s where the second-order effect lives, not in the 2.1-point spread, but in the RFPs it unlocks. Watch for procurement cycle data in Q2 earnings commentary from enterprise software vendors who sit adjacent to AI deployment.