OpenAI narrowed Anthropic's lead in Ramp business AI adoption to 3.2 percentage points in September
Ramp's September data has 43.8% of its businesses paying Anthropic and 40.6% paying OpenAI, a 3.2-point gap after OpenAI gained 1.12 points. The figures imply more than half of paying firms buy from both, so the gap says little about vendor lock-in.
The Engineer · Build desk
Key figures: at least 28.5% of Ramp businesses pay for both Anthropic and OpenAI. 55.9% pay for any AI. 43.8% pay Anthropic and 40.6% pay OpenAI, September figures.
| Figure | Value | Note | Claim |
|---|---|---|---|
| Pay both Anthropic and OpenAI, at least | 28.5% of businesses | 17 | |
| Pay for any AI | 55.9% of businesses | 4 | |
| Pay Anthropic, Sept. | 43.8% of businesses | 2 | |
| Pay OpenAI, Sept. | 40.6% of businesses | 3 |
What happened
- Overall AI adoption among Ramp businesses reached 55.9%, up 0.62 points, slightly faster than August's 0.40-point gain.
- The letter's author says growth in new AI customers has slowed significantly and that further market growth will have to come from more intense adoption.
- At the top 1% of firms, monthly spend per employee was $9,568 in September, 3.1% below the July high of $9,873.
- xAI rose 0.6 points on the popularity of Grok Bot, while Mistral AI sits below 1% adoption among U.S. firms.
Why it matters
- Who has to decide decision A second model vendor is already an ordinary budget line, so picking between the two on the adoption gap has thin support in this data.
- What it sets up precedent If labs concede discounts off sticker price, list prices overstate what contract customers pay, and vendor comparisons need negotiated rates.
- Who's exposed exposure Any benchmark built on the top-1% series inherits an estimate that is volatile, was revised upward, and may change again.
The gap is 3.2 points [14]. It shrank by 0.87 points in a month, because OpenAI gained 1.12 points and Anthropic gained 0.25 [15]. Anthropic's adoption rose too [2]. OpenAI's rose faster [3]. The letter reports one month of change, and one month is thin ground for a trend.
The letter's author is wary of the metric. "I sometimes question the value of this metric," the author wrote [5]. The stated reason is that more than half of businesses now pay for AI and many of them pay for several models [6].
The overlap can be bounded. Every business paying either vendor sits inside the 55.9% overall figure [4], so at least 43.8 + 40.6 - 55.9 = 28.5% of Ramp businesses pay both [17]. That is more than half of the businesses paying for any AI [18]. The two columns sum to 84.4% [16] in a market where 55.9% pay for anything [4].
For the gap to transfer to your budget, your firm would have to count the way the metric counts. The metric counts businesses. Dollars sit elsewhere: the letter says 80% of spend on OpenAI and Anthropic is driven by 1% of firms [20].
Spend itself is moving differently from adoption. Token volumes rose while token spend showed virtually no growth across August and September [11]. Spend then reached a new peak at the start of October, 1.5% above August's, and the letter still describes a significant slowdown in growth [12]. In the author's view, the main driver is price declines at the frontier, plus efficiency and quality gains across model tiers, pulling the blended price per million tokens down [19].
On lock-in, the letter's evidence is secondhand. The author bases it on what the author has heard from businesses that signed or saw the deals [21]. Those businesses describe contracts that rarely assume higher-than-normal spend, and labs that offer discounts off sticker price when they do [21]. "The contracts are not particularly restrictive," the author wrote [22]. The author adds that "These contracts could contribute to price cuts without driving additional volume" [23]. The letter also links a Ramp post by Ryan Stevens on how it is getting easier for companies to switch between models [13].
We think a vendor budget should start from negotiated price per million tokens and the cost of running two vendors, with the adoption gap as background. The context: the pricing evidence in this letter is the author's opinion and secondhand reports. A team can test that direction in its own negotiation.
What to watch
- Ramp's next monthly AI Index update: whether OpenAI's gain on Anthropic holds beyond one month.
- Whether Ramp revises the top-1% per-employee series again as it adds AI vendors to its estimates.
- Whether token spend resumes growth after two flat months, or the October peak proves to be a one-month rise.
Clarity's read
What the record supports and how the coverage leans. The claims behind it follow.
Reality
- Evidence62
- Adoption56
- Hype gap−10
- Incentives40
- Confidence65
Claim ledger
Ranked by verification strength, evidence, and original report placement.
- [1]
In September, OpenAI narrowed Anthropic's lead in business AI adoption among Ramp businesses.
- [2]
Among Ramp businesses, 43.8% paid Anthropic in September, up 0.25 percentage points from the previous month.
- [3]
OpenAI adoption among Ramp businesses rose 1.12 percentage points to 40.6% in September, outpacing overall AI adoption.
- [4]
Overall AI adoption among Ramp businesses reached 55.9%, up 0.62 percentage points, slightly faster than August's 0.40 point gain.
- [5]
The letter's author wrote: "I sometimes question the value of this metric."
- [6]
The author says the metric mattered when AI was new and OpenAI dominated, but today more than half of businesses pay for AI and many of them pay for multiple models.
- [7]
The author says growth in new AI customers has slowed significantly and that it feels like the top of the S-curve, where market growth will have to come from more intense adoption rather than more customers.
- [8]
xAI is a small player but rose 0.6 percentage points on the popularity of Grok Bot, while Mistral AI lags the market at less than 1% adoption by U.S. firms.
- [9]
The top 1% of firms spent $9,568 per employee per month in September, down 3.1% from their July high of $9,873, although per-employee spend rose 11.7% from the prior month.
- [10]
Ramp's methodological note says the top-1% spending estimate is more volatile than the median and top-10% estimates, was revised up across cohorts in all periods after new AI vendors were added, and is subject to change.
- [11]
Token volumes rose over the same period; token spend led the declines, and August and September produced two straight months of virtually no growth in token spend.
- [12]
AI spend rose at the beginning of October to a new peak, 1.5% above the previous peak in August, but the letter says a significant slowdown in growth remains.
- [13]
The letter links a Ramp post, credited to Ryan Stevens, on how it is getting easier and easier for companies to switch between models.
- [14]
Anthropic leads OpenAI by 3.2 percentage points in September paid adoption.
- [15]
The Anthropic-OpenAI gap narrowed by 0.87 percentage points in September.
- [16]
The Anthropic and OpenAI adoption figures sum to 84.4%.
- [17]
At least 28.5% of Ramp businesses pay for both Anthropic and OpenAI, since the union of the two cannot exceed the 55.9% paying for any AI.
- [18]
The minimum overlap is more than half of businesses paying for any AI (about 51%).
- [19]
In the author's view, the main driver is price declines at the frontier, in addition to efficiency and quality improvements across intelligence levels, driving the blended price per million tokens down.
- [20]
The letter says 80% of spend on OpenAI and Anthropic is driven by 1% of firms.
- [21]
Based on what the author has heard from businesses who signed or saw OpenAI and Anthropic deals, the customer is in the favorable negotiating position; contracts are rarely signed assuming higher-than-normal AI spend, and where they are, the labs offer discounts from sticker price.
- [22]
The author wrote: "In other words, the contracts are not particularly restrictive."
- [23]
The author wrote: "These contracts could contribute to price cuts without driving additional volume."
Get this ledger inside ChatGPT. Ask about any of our stories and the answer brings its figure, each claim’s verdict and the sources. Free, no account.
Sources
1 independent publisher whose own reporting we read for this story.
- ramp.combusiness spending data
1 article · October 16, 2026
Topics and entities
Follow any of these and your For You feed starts watching them — no settings page required.