Anthropic overtakes OpenAI in business adoption, despite government feud

Anthropic’s fight with the Trump administration may be doubling as a sales pitch. Ramp’s latest business-spend data show the San Francisco AI company reaching 34.4% of U.S. businesses in April, just ahead of OpenAI at 32.3%, after months of unusually public conflict with the White House and the Pentagon.

Ramp’s AI Index is built on transaction data from more than 70,000 U.S. businesses using its corporate card and bill-pay platform, and it tracks paid subscriptions to AI products. In Ramp’s March 2026 update, 24.4% of businesses were paying for Anthropic, up from about 1 in 25 a year earlier, while OpenAI’s adoption had slipped by 1.5 percentage points. Overall AI adoption among Ramp customers reached 50.6% in April, underscoring how quickly paid AI tools have moved into ordinary business spending.

The bigger question is whether Anthropic is being rewarded for the fight or for the product. Ramp economist Ara Kharazian has argued that Anthropic is especially strong in finance, tech and professional services, and that the company wins about 70% of head-to-head contests against OpenAI among first-time buyers of AI services. Ramp also said Anthropic quadrupled business adoption over the previous year, while OpenAI grew by just 0.3%, a gap that points to product pull, not just political theater. Anthropic has also said it is still constrained by compute and usage limits, which Ramp says may mean the company is turning away revenue.

The policy rupture has been stark. On February 24, Defense Secretary Pete Hegseth gave Anthropic chief executive Dario Amodei a deadline to allow unrestricted use of the company’s models for all legal purposes. Anthropic refused. Three days later, President Donald Trump ordered federal agencies to stop using Anthropic products and Hegseth designated the company a supply-chain risk. Anthropic sued on March 9, Judge Rita F. Lin granted a preliminary injunction on March 26, and a D.C. Circuit panel denied a stay on April 8, leaving the designation in place while the case moved forward.

The dispute has already split customers. Some defense-tech companies told employees to stop using Claude and switch to rival models after the blacklist, even as Anthropic continued talks with officials over the broader relationship. That makes the paradox sharper: the public clash may burnish Anthropic’s image as a vendor willing to set hard safety boundaries, but it also threatens a slice of government and defense business. For now, the numbers suggest enterprise demand and model performance are doing most of the work. The feud may help at the margins, but it is unlikely to explain a lead built across thousands of paying companies.