Enterprise AI spending is proving harder to lock down than vendors hoped. New data from Ramp, a corporate card and expense management platform (a tool businesses use to track and authorize company spending), shows Anthropic holding nearly 44% market share versus OpenAI’s nearly 40% as of July 2026. But OpenAI is growing faster in Q3 to date. The dataset covers more than 70,000 American businesses spending billions through Ramp’s bill pay and corporate card products.
How Did Anthropic Overtake OpenAI With Business Users?
OpenAI was the clear frontrunner in enterprise AI through early 2026. That changed in May, according to Ramp data published August 20, 2026. Anthropic crossed 41% market share that month while OpenAI dropped to 39%. OpenAI has not regained the lead since. As of July, Anthropic sits at nearly 44% versus OpenAI’s nearly 40%.
Ramp economist Ara Kharazian points to two factors driving the most recent dynamics. On OpenAI’s side, GPT-5.6 Sol has become “increasingly the choice for developers,” he posted on X. On Anthropic’s side, the Fable model tier (Anthropic’s higher-end offering, built for specific targeted use cases rather than general chatbot use) “disappointed both in adoption and real-world application given price + data retention requirements imposed by regulators,” Kharazian noted.
A separate source of friction: Anthropic drew criticism when it notified Fable users that the platform must retain their data for 30 days. For enterprise buyers with strict data governance requirements (internal rules about where company data can be stored and for how long), that kind of mandatory retention is a procurement blocker, not just an inconvenience.
Is OpenAI’s Q3 Growth Enough to Close the Enterprise AI Gap?
Not yet, and the outcome is genuinely uncertain. Ramp’s data shows OpenAI growing faster in Q3 2026 to date, but roughly a month remains in the quarter. Kharazian acknowledged this explicitly, framing the remaining period as “30 AI years” of time in which trends could easily reverse. He noted this in the context of cautioning against reading too much into the Q3 signal so far.
A few important caveats about the dataset:
- Ramp’s customer base skews toward technology companies, reflecting its roots as a popular Silicon Valley corporate card. Non-tech industries are underrepresented.
- Large enterprises using spend management tools from providers like American Express are excluded entirely.
- Ramp shared percentage breakdowns only. Actual dollar amounts were not disclosed.
- Both OpenAI and Anthropic are privately held and have not released audited financials, making third-party proxies like Ramp’s data among the few available signals.
The dataset is directionally useful but does not represent the total enterprise AI market.
What Does the Overall Enterprise AI Adoption Trend Show?
The structural shift underneath the head-to-head share battle may be more significant than the share split itself. According to Ramp data published August 2026, the percentage of its business customers paying for any AI tool topped 50% in March 2026 and reached nearly 56% by July 2026. That is consistent month-over-month growth across the sample.
This means both OpenAI and Anthropic are likely growing absolute revenue even as they compete for relative share. The market is expanding. Businesses switching between AI vendors reflects active experimentation, not a zero-sum game where one lab’s gain is the other’s loss.
Ramp’s data also suggests enterprise AI spending is not yet sticky. Businesses have shown they will move from one platform to another within a single quarter, often in response to a new model release or a change in pricing and policy.
What This Means for AI-Search Visibility
Here is the practical point: if businesses switch between AI tools this easily, the AI your customers are using today may not be the one they use next quarter. That matters for any brand trying to show up in AI-generated answers.
Think of it this way. A B2B software company has spent months ensuring Claude recommends them when enterprise buyers ask about tools in their category. But if those buyers migrate to ChatGPT, that positioning may not transfer. Visibility inside one AI model does not automatically carry over to another.
Hingewise’s read: the market share volatility Ramp is capturing is not just a competitive story between two AI labs. It is a signal that the AI tools generating answers for business users are still in flux. A brand visible on Claude today has no guarantee of the same visibility on GPT-5.6 Sol next quarter, or on whatever model gains traction after that. Building brand presence across multiple AI systems, rather than optimizing for one, reflects the reality the Ramp data describes.
The Ramp data also surfaces something the source does not explicitly flag. Compliance friction, specifically Anthropic’s 30-day data retention requirement for Fable, appears to have contributed to slower Fable adoption. This is a reminder that AI visibility is not only a content optimization question. Procurement decisions inside your target customers’ organizations shape which AI they use, and therefore which AI systems need to know your brand exists. Compliance policy changes at an AI vendor can shift your customers’ tooling overnight.
One additional caution: Ramp’s dataset skews heavily toward tech companies. Enterprise AI adoption rates and model preferences in healthcare, legal, financial services, or manufacturing may look quite different. Extrapolating these percentages across all industries would overstate their generalizability.
Before Your Team Commits to a Single AI Vendor Strategy
- Audit which AI tools your team actually uses day-to-day, not just which licenses you hold.
- Ask whether your target customers’ industries face data retention or compliance restrictions that limit which AI tools they can adopt.
- Check whether your content is structured for discoverability across multiple AI systems, not just one.
- Track which AI model version employees default to when options are available. Defaults shift with each major release.
- Monitor whether your customers’ AI tool preferences are changing. If enterprise AI spend is volatile at the macro level, it is likely volatile within your specific customer segments too.
- Note any policy changes from AI vendors (data retention, pricing tiers, access restrictions) that could affect enterprise procurement decisions.
The broader trend Ramp’s data confirms: no single AI vendor has locked in enterprise users. Model releases, pricing changes, and compliance policies are all moving fast enough to shift spending within a quarter. How durable enterprise AI brand loyalty turns out to be will become clearer once both OpenAI and Anthropic disclose actual financials ahead of their planned IPOs.
Source: TechCrunch, August 20, 2026. Ramp data covers U.S. businesses on the Ramp platform as of July 2026.
Lam Nguyen · Hingewise