OpenAI narrows Anthropic's enterprise lead as business AI spending proves volatile
The AMW Read
Ramp's business-spend data meaningfully updates the competitive standing of both frontier-lab case studies and shows enterprise switching costs are lower than assumed, without resolving the broader stickiness debate.
Named counterparties: Anthropic
OpenAI narrows Anthropic's enterprise lead as business AI spending proves volatile
New data from corporate card and expense platform Ramp shows OpenAI clawing back share of U.S. business AI spending after losing the lead to Anthropic in May. Tracking more than 70,000 American businesses, Ramp found Anthropic held roughly 41% of paying-customer share to OpenAI's 39% in May; by July Anthropic had grown to nearly 44% against OpenAI's nearly 40%, but Ramp economist Ara Kharazian says OpenAI is growing faster quarter-to-date in Q3. Ramp credits GPT-5.6 Sol's traction with developers, while Anthropic's higher-end Fable tier drew pushback over pricing and a 30-day data-retention requirement.
The reversal undercuts any assumption that enterprise AI spend, once won, stays won. With paying-customer share swinging several points in a single quarter, neither lab has a durable moat with business buyers who evaluate month to month — though the underlying market is still expanding fast, with the share of Ramp customers paying for AI climbing from just over 50% in March to nearly 56% in July, so both companies can grow revenue while trading share. It also lines up with OpenAI's own recent move to launch a no-retention Private Safety Processing feature explicitly positioned against Anthropic's data-retention policy — a direct product response to the friction Ramp's data now shows up in enterprise choice. Per the AI Market Watch index, OpenAI carries $199.6B in total funding to date (index coverage, not a full-market census), underscoring how much capital is riding on outcomes this fluid.
For investors evaluating either company ahead of a prospective IPO, the Ramp numbers argue against reading any single quarter's enterprise share as durable — retention policy, pricing, and model-release cadence each moved the needle within weeks. Builders selling into enterprises should expect procurement teams to keep dual- or multi-sourcing across model providers rather than standardizing on one, and should treat data-retention terms as a live competitive lever, not fine print.
