OpenAI is growing faster than Anthropic among U.S. businesses so far this quarter, according to new Ramp spending data posted August 20, 2026 by the expense-management company’s lead economist, Ara Kharazian, marking the first signs of a reversal since Anthropic took the lead in corporate AI spending three months ago.
The numbers come from the Ramp AI Index, a monthly measure of AI adoption and spend built from transaction data across more than 70,000 American businesses using Ramp’s corporate card and bill pay platform. Because it tracks paid receipts and invoices rather than survey responses, the index has become one of the few outside windows into how the two private AI labs’ enterprise businesses are actually performing.
The most recent full month of data, covering July 2026 and published August 12, 2026, shows Anthropic at 43.5% of U.S. businesses paying for its subscriptions or tokens, up 1.1 percentage points month over month, against OpenAI at 39.7%, up just 0.23 points. Anthropic has held the lead in recent months, extending it in July 2026 to 43.5% of businesses against OpenAI’s 39.7%, per Ramp’s data.
The quarter-to-date picture is what changed. Kharazian wrote that OpenAI is currently outgrowing Anthropic among Ramp’s business users in Q3 to date, crediting OpenAI’s newest flagship model. “GPT-5.6 Sol is really good, increasingly the choice for developers,” he posted. “Fable 5, meanwhile, disappointed both in adoption and real-world application given price + data retention requirements imposed by regulators.”
What the July Numbers Show
The model-level data in Ramp’s August update explains some of the shift. Fable 5, Anthropic’s top-tier model released in July 2026, accounted for only 6% of the tokens businesses purchased from Anthropic in its first month and 11.4% of dollars spent on Anthropic models, despite being the company’s most expensive offering at roughly $10 per million tokens. OpenAI’s GPT-5.6 Sol, by contrast, made up 25% of OpenAI tokens purchased and 23% of spend at half that price. In July, Fable 5 generated approximately 75% as much model-attributed spend as GPT-5.6 Sol, per Ramp’s token spend management data.
Kharazian’s post flagged price and data retention requirements imposed by regulators as factors in Fable 5’s slower uptake.
A Market Growing Fast Enough for Both
The rivalry is playing out on a rising tide. The Ramp AI Index shows overall paid AI adoption among its business customers reaching 55.7%, up from just over 50% in March 2026, and spending per customer climbing at every tier. In July, the median AI-spending business spent $11.95 per employee per month on AI, while the top 10% spent $650 and the top 1% spent a median of $7,400, according to Ramp’s data. AI spend on the platform has grown roughly fourfold over the past year.
There is also a third current worth watching: 6.1% of AI-spending businesses used model-serving or inference platforms, which route to open-source and Chinese-developed models, in July, up from 4.5% in January 2026. Kharazian notes that first-time AI buyers are still choosing the American labs almost exclusively, but the heaviest existing spenders are increasingly adding open-source options.
What the Ramp Data Does and Doesn’t Capture
The index comes with stated limits. Its sample skews toward the technology industry, reflecting Ramp’s customer base, and it excludes large enterprises whose AI spending runs through other expense-management providers. It also counts only paid transactions, so businesses using free AI tiers or employees expensing nothing on personal accounts go uncounted, meaning actual adoption is likely higher than the index shows. Ramp shares percentages rather than dollar totals, and the model-level figures come from a subset of customers using its token spend management product, a sample Ramp says skews slightly more tech-heavy than the broader index.
With a month left in the quarter, the Q3 growth figures are a snapshot, not a verdict. The next full Ramp AI Index update, covering August 2026 spending, will show whether OpenAI’s current pace was enough to retake the lead it lost in the spring.
