Anthropic Tops OpenAI in Business Customer Share
Key insights
- Anthropic reached 34.4% business customer share in April 2026, edging past OpenAI's 32.3% for the first time.
- Anthropic gained 3.8 percentage points in April alone while OpenAI shed 2.9 points in the same period.
- Ramp's expense data tracks actual corporate payments, making it a harder adoption signal than survey-based market share estimates.
Why this matters
Enterprise wallet share is a leading indicator of where API standardization settles -- teams that pay for Anthropic today are building workflows, evals, and integrations that create switching costs tomorrow. A single-month swing of nearly 7 combined points suggests active migration, not just new-customer distribution, which means OpenAI is losing accounts it already held. For founders and technical leaders choosing a primary model provider, this data shifts the calculus on long-term vendor risk and negotiating leverage.
Summary
Anthropic has displaced OpenAI at the top of the enterprise AI market for the first time, according to expense data from fintech platform Ramp. Among businesses tracked by Ramp, 34.4% are now paying for Anthropic services versus 32.3% for OpenAI -- a reversal that would have seemed unlikely even six months ago.
The shift accelerated sharply in April, with Anthropic gaining 3.8 percentage points in a single month while OpenAI dropped 2.9 points. Ramp's dataset covers real corporate spend, not survey responses, which makes it a harder signal than typical adoption polls.
Essentially: (Anthropic, OpenAI) the enterprise layer is now a contested market, not a default.
- April 2026 marks the first month any lab has held the top enterprise customer share position over OpenAI.
- Anthropic's gain of 3.8 points in one month suggests active switching, not just new entrants choosing Anthropic.
- The Ramp dataset reflects payment volume across participating businesses, so it captures committed spend rather than free-tier or trial usage.
OpenAI still leads on raw revenue and consumer mindshare, but enterprise wallet share is now a different story.
Potential risks and opportunities
Risks
- OpenAI faces accelerating enterprise churn if April's trajectory continues into Q3 2026, putting pressure on its reported $12B ARR run rate ahead of any IPO timeline.
- Anthropic's infrastructure could face capacity strain if enterprise onboarding accelerates faster than its AWS-backed compute agreements scale, degrading reliability for newly won accounts.
- Businesses that standardized deeply on OpenAI's API surface -- function calling schemas, Assistants API, fine-tuning pipelines -- face non-trivial migration costs if internal pressure pushes toward Anthropic, creating internal engineering friction and potential project delays.
Opportunities
- LLM observability and routing vendors (LangSmith, Braintrust, PortKey) benefit as enterprises managing multi-provider portfolios need tooling to track cost and quality across Anthropic and OpenAI simultaneously.
- Anthropic has leverage to accelerate enterprise contract negotiations with favorable pricing or SLA terms before OpenAI responds, locking in the cohort of April switchers before competitive retention efforts begin.
- Consulting and systems integration firms (Accenture, Deloitte AI practices) can position Anthropic migration assessments as a billable service line for OpenAI-heavy clients reassessing their primary provider.
What we don't know yet
- Whether the April swing was driven by a specific Anthropic release or pricing change, or by an OpenAI service disruption -- the Ramp data shows the outcome but not the trigger.
- How Ramp's participating business sample skews by company size and sector, and whether the share reversal holds across mid-market versus enterprise tiers specifically.
- Whether OpenAI's 2.9-point April drop reflects churn to Anthropic, churn to other providers like Google or Mistral, or a reduction in overall AI spend.
Originally reported by TechCrunch
Read the original article →Original headline: Anthropic Surpasses OpenAI in Business Customer Count for First Time, Per Ramp Expense Data — 34.4% vs. 32.3%