Anthropic now has more business customers than OpenAI, according to Ramp data

Anthropic now has 34.4% of participating Ramp businesses paying for its services, overtaking OpenAI at 32.3% for the first time. Ramp data show Anthropic’s share rose 26 percentage points over the past 12 months from 9% in May 2025, while OpenAI’s share fell 1 point. The article suggests Anthropic has gained share through strong execution with technical customers, though the advantage may not be durable.

Analysis

The key market signal is not that Anthropic is winning share, but that enterprise AI spend is concentrating around the model that best fits high-compliance workflows. That tends to favor vendors with stronger gross-margin durability because regulated buyers care less about sticker price and more about reliability, auditability, and refusal behavior. The second-order effect is pressure on the broader AI distribution stack: orchestration, evaluation, and workflow tooling should accrue more value than generic chatbot wrappers as customers standardize on one or two foundation models.

For Ramp, the data point is strategically useful even if not immediately monetizable. A business-expense dataset that can surface vendor-share trends becomes a proprietary intelligence product for finance teams and procurement, which raises the odds of incremental attach into analytics and spend-management workflows. The bigger implication is that AI adoption is no longer a single-vendor land grab; it is fragmenting by use case, which reduces the likelihood of one incumbent sustaining category dominance on brand alone.

The risk to the Anthropic outperformance narrative is that this is still a usage-intensity story, not a durable budget-share story. If OpenAI improves enterprise controls, bundles aggressively, or undercuts via distribution partners, share can swing back within 1-2 quarters; the moat here is execution, not switching costs. In the near term, the main catalyst is enterprise procurement renewal season over the next 1-3 quarters, where proof of productivity and compliance will matter more than model benchmarks.

Contrarian take: consensus may be overestimating the importance of model preference and underestimating channel power. The durable winners may be the platforms that sit between the buyer and the model, because customers increasingly want optionality rather than allegiance. That makes the revenue mix of any AI spend beneficiary more important than headline share leadership.

Key Decisions for Investors

• Long RAMP on a 3-6 month horizon into earnings/AI-product updates; the dataset itself is a monetizable asset, and any evidence of higher AI attach rates can expand multiple perception even if core spend management growth is unchanged. Risk/reward favors a medium-term re-rate if management frames this as a proprietary intelligence wedge.
• Pair trade: long AI workflow/orchestration beneficiaries vs short single-model exposure on 6-12 month horizon. Prefer companies monetizing model-agnostic automation over pure model vendors; the thesis is that buyer optionality compresses winner-take-all economics and shifts margin pool upstream in the stack.
• Buy 3-6 month calls on RAMP into potential enterprise data-product commentary, funded by selling OTM calls if implied vol is elevated. The setup is asymmetric if investors start valuing the firm as both a payments platform and an AI spend data source.
• Maintain a tactical hedge against OpenAI-related ecosystem beneficiaries that rely on brand momentum alone; if enterprise budget share is fragmenting, those names are vulnerable to multiple compression over the next 1-2 quarters.
• Watch for signs of procurement standardization and multi-model routing in finance/tech. If that accelerates, rotate away from foundation-model beta and into picks-and-shovels software with usage-based revenue tied to AI deployment rather than model selection.