Skip to content
NewsResearch

Anthropic leads OpenAI 44 to 40 in US enterprise AI spend, but Ramp data shows high switching rates as models cycle

· by Pondero Newsdesk

The short version

Spending data from Ramp's 70,000-plus US business customers shows Anthropic at roughly 44 percent enterprise AI market share versus OpenAI's 40 percent as of July 2026, with OpenAI growing faster in Q3 and high switching rates signaling low stickiness for both platforms.

Anthropic leads OpenAI 44 to 40 in US enterprise AI spend, but Ramp data shows high switching rates as models cycle

Corporate expense data from more than 70,000 US businesses shows Anthropic still holds the enterprise AI market share lead over OpenAI, but the gap is closing. The more significant finding in the August 20 Ramp report is what it reveals about commitment: US companies are switching AI vendors freely every time a major new model ships, a pattern that challenges the assumption that either lab is building durable enterprise loyalty.

What the data shows

Ramp, a corporate credit card and expense management platform, published spending figures covering more than 70,000 US businesses on August 20. As of July 2026, Anthropic held roughly 44 percent of AI market share among those customers versus OpenAI's roughly 40 percent, per TechCrunch's report on the Ramp data.

Anthropic first overtook OpenAI in May 2026, when it reached 41 percent against OpenAI's 39 percent, per the same report. OpenAI has not retaken the lead since. Ramp economist Ara Kharazian noted, however, that OpenAI is growing faster among this customer base in Q3 to date, with a full month of the quarter still remaining.

The total addressable pool of paying customers is also expanding. The share of Ramp businesses paying for any AI service topped 50 percent in March 2026 and climbed to nearly 56 percent by July, per the same report.

Ramp's data covers businesses that route spend through its bill-pay and corporate card products, which skew toward the tech industry given the platform's origins as a Silicon Valley fintech. Large enterprises using spend-management tools from providers such as American Express are not captured.

What drove the Q3 shift

Kharazian attributed OpenAI's Q3 momentum partly to GPT-5.6 Sol, which he described as "really good" and "increasingly the choice for developers," per his post on X. He cited two headwinds for Anthropic's Fable 5: its price point and a 30-day data retention requirement that Anthropic notified Fable customers about in August 2026, which generated criticism from enterprise users.

The privacy protections OpenAI announced the day before, on August 19, appear positioned to exploit that backlash. TechCrunch reported on the OpenAI move as a direct response to Anthropic's data retention announcement.

Kharazian acknowledged the picture is not simple. Fable 5 targets narrower use cases than a general-purpose chatbot, and the quarter is not finished.

Why it matters

For enterprise operators, the central practical implication of the Ramp data is that neither platform has secured the kind of sticky relationship that justifies a single-vendor lock-in strategy. TechCrunch's report described businesses as "willing to flop back and forth as each lab releases new models." That volatility works in procurement teams' favor when negotiating, but it also means that production systems built on a specific model version face real disruption risk as enterprise spending drifts to each successive release.

The Fable 5 data retention development is worth tracking separately from the market share figures. If a regulatory retention requirement measurably redirected enterprise spend, compliance and security teams at organizations already running Fable 5 workloads need to revisit their data governance posture, independent of whatever model they end up preferring long-term.

What to watch next

Ramp plans to publish updated spending data quarterly. The next report will show whether OpenAI's Q3 gains held through Anthropic's expected late-2026 model releases. The impact of OpenAI's new customer privacy protections on enterprise switching behavior in Q4 is also worth tracking.

Sources