A year ago, asking which AI company dominated enterprise spending was barely a question: OpenAI was the obvious answer. This week, new data from fintech firm Ramp showed that Anthropic has taken the lead, with 34.4% of surveyed businesses paying for Anthropic products in April versus 32.3% for OpenAI. It is the first time Anthropic has come out on top. The gap is narrow, the dataset covers 50,000 US businesses rather than the whole market, and these numbers will move again. But the direction of travel matters.
Ramp's AI Index tracks actual company spending on AI tools, not survey responses about intent. That makes it a more grounded measure than most. The index shows Anthropic climbing from 9% business adoption in May 2025 to roughly 35% a year later, a gain of 26 percentage points. OpenAI's share dipped slightly over the same period, even as total corporate spending on AI continued to rise. Both companies are growing in absolute terms; Anthropic is simply growing faster.
The engine of Anthropic's rise was Claude Code, its AI-assisted coding tool, which gained serious traction with developer and engineering teams in late 2025. Ramp's economist Ara Kharazian told TechCrunch that Anthropic has been ahead among high-adoption groups, including finance, tech, and professional services firms, for some time. The recent shift represents the gap closing in sectors where OpenAI previously had a stronger foothold.
The PwC deal announced this week is the clearest sign of where Anthropic is taking this momentum. Under the expanded partnership, PwC will train 30,000 of its professionals on Claude, with an eventual target of "hundreds of thousands" of employees globally. The two firms are also setting up a joint Centre of Excellence focused on enterprise AI deployment. Dario Amodei, Anthropic's CEO, described the collaboration in terms of concrete efficiency gains: insurance underwriting that took ten weeks now takes ten days, security work that took hours now takes minutes.
These are the claims consultancies have been making about software for forty years, so some scepticism is warranted. But the scale of the commitment signals something real: major professional services firms are now training their entire workforce on a specific AI provider's tools, not running isolated pilots. That creates switching costs and workflow dependencies that are difficult to unwind, regardless of which model happens to score best on a benchmark next quarter.
OpenAI has not responded by staying still. This week the company launched what was described as a dedicated enterprise deployment division, backed by more than four billion dollars and staffed with engineers who embed directly with corporate clients to help integrate AI into operations. OpenAI is repositioning itself from a model provider into something closer to an infrastructure and consulting company, a move that echoes how cloud providers evolved from "here are servers" to "we will run your entire IT stack."
The real story in the Ramp data may not be which company is ahead, but what the shift reveals about how enterprise AI decisions are actually being made. Price, reliability, and tool fit are increasingly the deciding factors, not brand recognition or benchmark performance. The AI market is beginning to resemble a mature software market, where buyers shop around, negotiate on terms, and switch providers when a better option emerges. That is a very different environment from the one OpenAI operated in when it was the only credible option at scale.
A 2.1-point lead in a monthly spending survey is not a moat. But twelve months of consistent trajectory, combined with infrastructure-level partnerships like the PwC deal, suggests Anthropic has moved from underdog challenger to co-leader in a way that is increasingly hard to dismiss as a temporary blip.