For the First Time, More U.S. Businesses Pay for Claude Than ChatGPT
The numbers have been converging for months. In April, they finally crossed.
Anthropic's share of U.S. business adoption rose 3.8 percentage points last month to 34.4%, while OpenAI's fell 2.9 points to 32.3%, according to the May 2026 edition of the Ramp AI Index — a monthly report tracking AI spending across more than 50,000 American companies. It marks the first time in the index's history that more U.S. businesses are paying for Claude than for ChatGPT.
The crossover is striking not just for what it is, but for how fast it arrived. In June 2023, Anthropic held a 0.03% share of business AI adoption. By April 2025, that had climbed to 7.94%. One year later: 34.44%. OpenAI, meanwhile, peaked at roughly 36.5% sometime in mid-2025 and has been on a slow slide since.
Claude Code Is Pulling the Weight
The proximate cause of Anthropic's surge is not hard to identify. Claude Code — the company's agentic coding tool — has become its fastest-growing product by a wide margin, and the ripple effects are now showing up in Ramp's payment data in sectors well beyond engineering.
One widely-cited estimate put Claude Code's footprint at approximately 4% of all public GitHub commits as of spring 2026, roughly double its share from the prior month. A separate Anthropic disclosure noted that 80% of the company's own new production code is now authored by Claude. The run rate for Claude Code alone has crossed $2.5 billion — built in under a year.
The dynamic is classic bottom-up enterprise penetration. Developers adopt the tool, productivity gains become visible, and budget follows. According to Ramp's economist Ara Kharazian, that pattern has been especially pronounced in high-value verticals. "Anthropic has already been in the lead amongst the high adoption groups like finance, tech, professional services," Kharazian told TechCrunch, while noting that "it's across the other firms where OpenAI still has a lead, but that has been shrinking over the past couple of months."
Overall AI adoption across the Ramp index reached 50.6% in April — meaning more than half of U.S. businesses tracked by the platform are now paying for at least one AI product.
The Other Side of the Ledger
Before Anthropic's win gets fully priced in, it's worth reading the fine print — including in the same Ramp report that declared the milestone.
The IDC's March 2026 survey of more than 1,000 enterprise organizations offers a different lens. Where Ramp measures breadth — which businesses are paying — IDC measures depth. On that metric, only 19% of organizations report extensive use of Claude, trailing both OpenAI and Google (at roughly 42% and 38%, respectively). Another 25% of respondents said they are actively evaluating Claude, suggesting real pipeline but not yet converted usage.
The distinction matters. A business that runs a few API calls through Claude each month shows up as "adopting" in Ramp's data. A business that has woven a model into its core workflows looks very different in the IDC survey. Anthropic is winning on the first measure; it is still catching up on the second.
Three Threats VentureBeat Flagged — and Why They're Real
VentureBeat's analysis of the Ramp data identified three structural vulnerabilities that could erode Anthropic's new position:
Token-based pricing incentives. Anthropic's revenue model rewards token consumption, which means the company benefits financially when users run expensive, high-volume queries — even when lighter models would do. Critics argue this misaligns incentives between Anthropic and the cost-conscious enterprise buyers it is trying to win.
Compute capacity constraints. Explosive growth in Claude Code usage has strained infrastructure. Users have reported rate limits, service degradation, and reliability issues at scale. For large enterprises making long-term bets on a vendor, uptime consistency often matters more than model quality at the margin.
Escalating costs from model updates. A recent Anthropic model update reportedly tripled token costs for prompts that include images — a significant price change that landed without much warning on customers who had budgeted around earlier rates.
None of these are fatal. But each is the kind of friction that gives enterprise procurement teams a reason to hedge, and gives OpenAI or Google a second look.
Why This Moment Is Still Significant
Even with the caveats, the Ramp crossover is a meaningful inflection point. For most of the generative AI era, OpenAI has been the default enterprise answer — the name in the budget line, the model in the procurement deck. That default is no longer automatic.
The trajectory Anthropic has traced — from 0.03% to 34.44% in less than three years — is the kind of growth curve that reshapes industries, not just competitive rankings. And Claude Code's role in driving it suggests that the next battleground for enterprise AI market share runs directly through the developer tools ecosystem: whoever owns the commit, often owns the contract.
What to Watch
The June Ramp AI Index — due next month — will show whether April's crossover was a durable shift or a single-month statistical artifact. Watch for whether Anthropic can hold above 33% while OpenAI stabilizes its decline, and whether the developer beachhead translates into deeper organizational adoption of the kind IDC's survey currently shows it lacks. Also worth tracking: any pricing or reliability announcements from Anthropic that address the compute and cost concerns flagged this month. A sustained lead will require more than momentum — it will require infrastructure that scales without surprising its customers.
"Anthropic has already been in the lead amongst the high adoption groups like finance, tech, professional services — it's across the other firms where OpenAI still has a lead, but that has been shrinking over the past couple of months."- Ara Kharazian, Economist, Ramp