📰 Key Takeaways

OpenAI is mounting a comeback against Anthropic in the US enterprise market. Fresh data from corporate card and expense management company Ramp shows Anthropic first overtook OpenAI in market share among its paying enterprise users in May, hitting 41% versus 39%; by July the gap widened further to nearly 44% versus nearly 40%, and OpenAI still hasn’t reclaimed the lead. That said, Ramp economist Ara Kharazian notes that looking at the latest quarter-to-date numbers, OpenAI is growing faster than Anthropic — and with a month left in the quarter, things could still flip again. Ramp only discloses percentages, not actual spending figures. The dataset covers more than 70,000 US businesses spending billions of dollars through Ramp’s bill pay and corporate card products. The customer base spans multiple industries but skews toward tech, since Ramp’s card is popular in Silicon Valley, and it doesn’t capture large enterprises using other financial tools like American Express — so this reflects a market trend rather than the full picture. Kharazian attributes it to OpenAI’s GPT-5.6 Sol performing well and increasingly becoming developers’ go-to choice, while Anthropic’s higher-tier Fable 5 model has underperformed in adoption and real-world use due to its pricing and a regulatory-driven 30-day data retention requirement — a policy that has previously drawn user complaints. Overall, the share of Ramp customers paying for AI keeps climbing: it first topped 50% in March and is now nearing 56% in July, showing the overall market pie is still expanding and both companies stand to grow with it. But the way enterprise customers keep swinging back and forth with each new model release is a reminder that “stickiness” in enterprise AI spend is still a variable investors need to watch.

💬 JudyAI Lab Take

Ramp’s data on more than 70,000 US corporate cards shows Anthropic’s enterprise paid market share first overtaking OpenAI in May (41% vs 39%), with the gap widening to nearly 44% vs nearly 40% by July — OpenAI still hasn’t taken it back.

This tug-of-war is a reminder that enterprise model choice isn’t just about who’s “stronger.” Ramp’s economist points out that OpenAI’s GPT-5.6 Sol has been performing well and is increasingly becoming developers’ default pick, while Anthropic’s higher-tier Fable 5 has underwhelmed on adoption due to pricing and its 30-day data retention policy. Turns out pricing and data governance can matter just as much as raw model capability. The share of Ramp customers paying for AI keeps rising — past 50% in March, nearly 56% in July — so the market itself is growing, but rankings keep swinging with every new model release, and that stickiness is still shaky.

For teams currently evaluating AI tools, it’s worth weighing pricing and data policy alongside performance when making the call.

📅 Source Info

🔗 Further Reading