Anthropic Overtakes OpenAI Among Business Customers

For the first time since the chatbot era began, more American businesses are paying for Anthropic than for OpenAI. The shift comes from the May 2026 release of the Ramp AI Index, which tracks corporate-card spending across roughly 40,000 firms and offers one of the more honest reads on enterprise adoption.

Anthropic's share rose 3.8 percentage points in April to 34.4% of businesses, while OpenAI's fell 2.9 points to 32.3%. Total AI spending across the index also nudged up to 50.6%, a sign the pie is still growing even as the slices change shape.

The pace of the swing is the part worth pausing on. Anthropic was at 0.03% of businesses in mid-2023 and only 7.94% a year ago. To go from there to a 34% lead in twelve months is the kind of curve OpenAI itself enjoyed during the ChatGPT moment. As Axios noted, OpenAI's enterprise share has been drifting downward since peaking near 36.5% in mid-2025.

The Claude Code effect

Most analyses point to one product as the engine of the shift. Claude Code, Anthropic's command-line coding agent, has become the company's fastest-growing release. TechCrunch reported that an analysis circulating among investors estimated 4% of all public commits on GitHub were authored by Claude Code, double the share from a month earlier.

That trajectory matters because coding is where enterprise budgets break loose first. Once a developer team standardises on a model, procurement follows quickly: API credits, team seats, and eventually the wider knowledge-worker rollout. OpenAI's recent moves, including its new $4 billion deployment unit and a non-exclusive Microsoft partnership, suggest the company knows this dynamic and is racing to defend the high end.

Where the lead could erode

The Ramp report does not read as a victory lap. As VentureBeat summarised, three problems hang over the win.

First, the revenue mix. Anthropic earns more when customers run the most expensive models, which gives the company an incentive to keep nudging users toward the Opus tier even when a cheaper or faster model would do the job. Buyers notice that kind of nudging, especially on renewal.

Second, reliability. Outages, rate limits, and capacity squeezes have been common complaints from heavy users this year. An enterprise that has reorganised its workflow around a coding agent does not look kindly on a Tuesday afternoon outage.

Third, output drift. Some enterprise teams are reporting that Claude's responses feel less consistent than they did six months ago, an effect that is hard to verify from the outside but easy to act on for buyers whose contracts are up for renewal.

The valuation question

Behind the adoption number is a fundraising story. Anthropic is reportedly in talks for a round that would value the company between $400 billion and $950 billion, as The AI Insider reported this week. That is not far short of OpenAI's last private valuation, and would put both firms in a category previously reserved for incumbent tech giants.

For now, the takeaway is simpler. Enterprise AI has stopped being a two-horse race won by default. The customers are choosing, and one of them, for the first time, has chosen Anthropic.