OpenAI Is Gaining on Anthropic With Business Users, Ramp Data Shows

Spend data from 70,000 companies shows the enterprise AI race is far from settled — and loyalty is thin

Published: 2026-08-22 Category: Quick Take Sources: TechCrunch

The story

With both OpenAI and Anthropic heading toward IPOs but not yet releasing financials, outsiders have to look elsewhere for signals. One such source is Ramp, the corporate card and expense company, which tracks AI spend across more than 70,000 U.S. businesses. Its new data shows OpenAI is starting to claw back ground from Anthropic.

OpenAI was once the runaway leader with business customers, but lost the top spot in May, when Anthropic hit 41% market share to OpenAI's 39%. OpenAI never regained the lead. As of July, Anthropic held nearly 44% to OpenAI's nearly 40%. Yet Ramp economist Ara Kharazian says OpenAI is currently growing faster among this segment in Q3 to date. Ramp's customers skew tech, and it declined to share actual dollar figures, only percentages.

The analysis

This is a reminder that the enterprise AI market is not sticky. Businesses are willing to flip-flop between labs as each releases new models — a volatility that should give both companies' future investors pause about how durable enterprise AI spend really is. Kharazian credits GPT-5.6 Sol's strength among developers, while noting Fable 5 "disappointed both in adoption and real-world application given price + data retention requirements." Anthropic did spark backlash by warning Fable users it must retain their data for 30 days.

The nuance: Fable is Anthropic's higher-end tier, built for targeted use cases rather than general chat — so a broad market-share figure can over-simplify. But the direction matters. Both companies are swimming in a growing tide: the share of Ramp customers paying for AI topped 50% in March and reached nearly 56% by July.

The analysis

Three takeaways. First, the pie is expanding, so both labs can grow revenue even while trading market share — this isn't a zero-sum fight yet. Second, spend is increasingly a function of each quarterly model release cycle: a strong new model wins customers, a weak or expensive one loses them. That's a brutal treadmill for pricing and product teams. Third, enterprise buyers treat frontier labs as interchangeable suppliers, not lock-in partners — a reality that should temper any investor assumptions about durable dominance. Watch the Q3 close: with a month left, "that's like 30 AI years," as Kharazian put it.

Reporting from TechCrunch's Julie Bort (August 20, 2026).