Corporate spending data reveals OpenAI clawing back ground lost to Anthropic since May

Aug 210 views

Key takeaways

Anthropic leads OpenAI among Ramp business users 44% to 40% as of July, but OpenAI is growing faster in Q3.
Ramp's dataset covers 70,000+ U.S. companies; it skews tech-heavy and excludes large enterprises using other spend tools.
The share of Ramp customers paying for AI at all rose from just above 50% in March to nearly 56% by July.

OpenAI is posting its fastest business-customer growth relative to Anthropic in months, according to new spending data released by Ramp, the corporate credit card and expense management platform used by more than 70,000 U.S. companies. The figures show OpenAI holding nearly 40% market share among Ramp's paying business users in July, while Anthropic sits at nearly 44% — but OpenAI's quarter-to-date growth rate in Q3 is currently outpacing its rival. Ramp economist Ara Kharazian attributed part of OpenAI's momentum to GPT-5.6 Sol, which he described on X as increasingly popular with developers.

The competitive dynamic looked very different just a few months ago. Anthropic first overtook OpenAI among Ramp's business customers in May, when it hit 41% market share compared to OpenAI's 39% — a milestone that marked the first time ChatGPT's parent had lost the enterprise lead. The gap widened through summer, and as of July Anthropic's advantage remains meaningful even as OpenAI's trajectory improves.

Kharazian also pointed to headwinds for Anthropic's higher-end Fable model tier, citing its premium price point and a regulatory-driven requirement that forces Anthropic to retain user data for 30 days — a policy that drew criticism when it was disclosed. Fable is designed for targeted enterprise use cases rather than general-purpose chat, which limits its addressable market but also means adoption metrics may not tell the full story of its commercial value.

Ramp's dataset is notable for its scale, but it carries important caveats. The platform skews toward tech-sector companies, and large enterprises that route spending through tools from American Express and similar providers are excluded entirely. Ramp also shared only percentages, not absolute dollar figures, which means the relative share movements can look dramatic even when underlying spend shifts are modest.

What the data does make clear, beyond the head-to-head race, is that the overall enterprise AI market keeps expanding. The share of Ramp customers paying for AI at all climbed above 50% in March and reached nearly 56% by July. That means both OpenAI and Anthropic are likely growing their business revenue even while trading blows over market share — though the ease with which enterprise customers switch allegiances as new models drop raises real questions about the long-term stickiness of AI spending.

The bigger picture

The most pointed takeaway from Ramp's numbers isn't who is winning today — it's how quickly the rankings change. Enterprise software markets typically exhibit high switching costs and slow churn; customers sign multi-year contracts, train internal workflows around specific tools, and resist migration. AI assistants, at least at this stage, appear to behave differently. Businesses seem willing to redirect spend within a single quarter based on model quality alone, which is an unusual dynamic that should concern anyone valuing these companies on the assumption of durable enterprise lock-in.

For rivals further down the stack — Google's Gemini for Workspace, Microsoft's Copilot suite, and Amazon's Bedrock offerings — this volatility cuts both ways. On one hand, it signals that OpenAI and Anthropic haven't yet built moats deep enough to freeze out competition. On the other hand, those hyperscalers have distribution advantages baked into products businesses already pay for, which may prove more resilient than raw model performance over time. If Anthropic's Fable tier is losing ground partly because of data-retention rules imposed by regulators, that's a sign that policy risk is now a real commercial variable, not just a background concern.

With both OpenAI and Anthropic still privately held and not required to disclose financials, third-party proxies like Ramp's data are among the few windows investors and analysts have into relative momentum. That makes Kharazian's quarterly reads worth tracking closely — not as precise revenue reporting, but as a leading indicator of which lab's product decisions are resonating with paying customers. The next model releases from either company could flip the chart again before Q3 even closes.

LagPing's take

We decided to cover Ramp's data because it's one of the rare concrete, third-party signals in a space where both major players guard their financials carefully. OpenAI and Anthropic both talk confidently about enterprise traction, but they're not showing receipts yet — Ramp is as close to receipts as we're likely to get until IPO filings arrive. What struck us most here isn't the percentage gap itself, which is fairly narrow, but the speed of the swings. We're used to writing about enterprise software as a slow-moving, contract-heavy world. AI is clearly playing by different rules right now, and that matters for how the whole industry gets valued and covered. We'll keep watching Ramp's quarterly reads as a barometer alongside model release cycles — the two seem tightly linked.

X / Twitter WhatsApp

View at TechCrunch AI

Find "OpenAI" on Amazon

As an Amazon Associate, LagPing earns from qualifying purchases. Product links are affiliate links.