AI Token Spending Stalls at Ramp While Usage Keeps Climbing

AI token spending among Ramp's business customers grew just 1.5% across August and September, after rising 42% month over month in July, according to TechCrunch's report of November 6, 2026. Over the same stretch, token volumes kept rising. Dollars stalled while consumption did not.

A token is the unit AI models bill by, roughly a word or fragment of a word in a prompt or response. That divergence is the central fact in the data. Spend and volume moved apart in the late-summer months, and the report attributes the gap to pricing as much as to usage. OpenAI cut its API prices in half in September for GPT-6 Sol and GPT-6 Luna. Anthropic cut pricing for Fable 5.1 and Opus 5.5, and said in a recent filing that the decision reduced its revenue growth by about $600 million.

Think of a fuel station that halves its price per gallon. Drivers can go twice as far for the same money, so a monthly fuel bill can stay flat even as the car is on the road more. A halving of unit price does not reduce consumption by the same factor, so a rising token count can coexist with flat billings. Both price changes fell inside the window where spend stalled.

Spending then re-accelerated. Ramp data cited by TechCrunch shows token spend rose 8.9% from September 23 to October 23. Ramp's October 2026 AI Index, titled "Cracks in the AI thesis, part 3," reported that the top 1% of firms spent $9,568 per employee, down 3.1% from its July high. TechCrunch's figures put that top cohort just below $10,000 per employee in August and September, also below summer peaks.

The distribution is steep. Per TechCrunch, median firms spend $12 to $15 per employee per month on AI, while the top 10% spend $600 to $1,000. Ramp lead economist Ara Kharazian told TechCrunch that almost the entire spend on AI comes from venture-backed, high-growth technology companies. Separately, Ramp data reported by Yahoo Finance on October 21 found that 1% of firms drive 80% of AI spending on OpenAI and Anthropic. Reuters, citing Ramp data for a week in October, reported that Anthropic accounted for 60.1% of AI token spending in that week. Ramp notes that its data captures only a slice of business spending.

Adoption breadth is moving on a different track from depth. Ramp's AI Index reported that 46.6% of businesses were using AI in its January 2026 edition. For September 2026, the index recorded 55.9% of U.S. businesses on Ramp paying for AI, up 0.6 points from August. The index draws on card and bill-pay data from more than 85,000 firms. In June 2025, the same index had shown U.S. business adoption leveling off at 41%. Ramp's Token Spend Intelligence product pulls token-level usage directly from providers including Anthropic and OpenAI, which is how the firm can separate volume from dollars.

Anthropic's own commitments are on a far larger scale. Its IPO filing, reported by Bloomberg on October 27, 2026, disclosed roughly $417 billion in future minimum payments, primarily for compute and data center infrastructure. As of November 2, 2026, Bloomberg reported that Anthropic had committed to $895 billion in spending as its IPO neared.

In my view, the number that matters most for the next few quarters is the one the headlines underplay: the share of spend held by a small group of firms. If 1% of customers account for roughly 80% of spend, the total can swing on a few procurement decisions, and a provider's revenue trajectory depends on how few budgets respond to price and capacity changes.

The broader context here is that the same price cuts that stalled the dollar line also widened what a firm can afford to run. The median figure of $12 to $15 per employee per month is small next to the top cohort's spend, and the volume data suggests usage is still expanding beneath the flat billings. Whether that broader base grows into spending on the scale of the top 1% of firms is an open question. Cheaper tokens are being consumed at a rising rate, and the people deciding how much of that consumption to pay for are, for now, concentrated in a narrow set of high-growth companies.