Claude AI Statistics 2026: Revenue, Users and Which to Cite
Claude statistics for 2026: Anthropic's S-1 reports $17.0 billion Q3 revenue and a $104.0 billion nine-month net loss, but no user count. Which number to cite.

On this page
Last verified: November 5, 2026.
Claude statistics are easy to find and hard to quote, because the figures look comparable and count different things. A run rate is not revenue, a visit is not a person, and a share needs a denominator. Marketer adoption surveys have the same base problem, sorted out in ai in marketing statistics.
This page gives each of the Claude AI statistics its measure, date and primary source, from revenue and profit to users and market share. It also shows where credible sources disagree and which figure to cite for which claim. If you have to defend a number in a deck or an article, start with the cite-this-for-that table near the end.
Claude statistics at a glance
Claude statistics fall into four kinds that count different things: filed revenue from Anthropic's S-1, run rates Anthropic announces, audience measures from Similarweb and Ramp, and Economic Index usage samples.
For the cross-engine referral side of AI search, see AI SEO statistics.
These are the ten key numbers, each with its measure, date and source:
- Anthropic's S-1 amendment reports revenue of $17.0 billion for the three months ended September 30, 2026, about $5.7 billion a month.
- Nine months of 2026 revenue were $33.2 billion, 7.2 times all of 2025's $4.6 billion.
- Operating income was $974 million in the second quarter and an operating loss of $567 million in the third.
- The nine-month net loss was $104.0 billion, of which $94.3 billion is a non-cash fair-value change.
- Anthropic said its run-rate revenue crossed $47 billion in May 2026; three anonymous sources put it at $65 billion at the end of July, and the S-1/A leaves the October 31 figure blank.
- More than 1,900 customers had $1 million or more of trailing twelve month revenue on September 30, 2026.
- Anthropic's Series H valued it at $965 billion post-money on May 28, 2026.
- Similarweb counted 103.9 million Claude app monthly active users in May 2026 and 950 million to 969 million claude.ai visits a month.
- Claude's share runs from 9.6% of chatbot web visits (Similarweb, August 2026) to 60.5% of business AI token spending (Ramp, week ending October 23, 2026).
- The S-1/A reports no user total; the 312.4 million figure that circulates is one vendor's model.
Last verified: November 5, 2026. Revenue is now filed, users are not, and every other figure needs its measure beside it.
Claude revenue in 2026: quarterly figures from the S-1
Claude revenue was $17.0 billion in the third quarter of 2026, about $5.7 billion a month, according to Anthropic's S-1 amendment filed November 2, 2026.
The Claude revenue 2026 figures come from two filings: Q2 and the six-month numbers from the original Anthropic S-1 of October 26, Q3 and the nine-month numbers from the amendment. The 2026 interim figures are unaudited.
Claude revenue by quarter and month
Recognized revenue grew about 2.5 times from Q1 to Q2 and about 47% from Q2 to Q3. Q1 is not reported on its own; we got it by subtracting Q2 from the six-month figure ($16.3 billion minus $11.6 billion).
Download CSV (CC BY 4.0)
The 2025 bar is the S-1/A's $4.6 billion over four quarters, shown as a quarterly average.
Revenue booked from 2023 through September 30, 2026 adds up to about $38.2 billion by our arithmetic (the S-1/A gives $41 million for 2023 and $386 million for 2024). Nearly all of it arrived in 2025 and 2026.
Anthropic run-rate revenue: what the number counts
The S-1 defines Run-Rate Revenue as the trailing 28 days of consumption fees multiplied by 13, plus monthly subscription fees multiplied by 12.
It moves with the last four weeks, and the filing says it is not a substitute for GAAP revenue and not a forecast.
Consumption-based revenue was 79% ($26.3 billion) of the nine months to September 30, 2026 and subscriptions 21% ($6.9 billion), so most of the metric rides on a 28-day consumption window.
That is why a run rate and a quarter differ. Here is every run-rate point Anthropic has announced, with the one number it left blank:
Download CSV (CC BY 4.0)
The sources are Anthropic's Series F, Series G and Series H announcements, its April 2026 compute announcement, and CNBC's report for July. Between about $9 billion at the close of 2025 and over $47 billion in May 2026, the run rate grew 5.2 times.
The filed quarters let you test those run rates:
- Q2 revenue times four is $46 billion, which is 99% of the $47 billion May run rate.
- Q3 revenue times four is $68 billion, 5% above the $65 billion July run rate.
That fits a run rate that kept rising through August and September, although the S-1/A leaves the October 31 figure empty and we infer none. This is a check, not a restatement: GAAP revenue and Run-Rate Revenue are different measures.
Claude revenue in 2025: the bounds and the filing
The filing puts 2025 revenue at $4.6 billion, and the run-rate points bracket it.
A lower bound of $2.3 billion holds each announced level until the next one; an upper bound of $6.3 billion assumes each level was already in place at the start of its interval. The filed figure sits inside both.
The S-1/A reports Anthropic revenue of $41 million, $386 million and $4.6 billion for 2023, 2024 and 2025. CNBC wrote in August that the company generated roughly $10 billion in revenue for all of 2025. That figure is 2.2 times the filed $4.6 billion, so it fails the filing test, and it probably mixed in a run rate.
Claude profit and loss in 2026
Claude profit depends on the basis.
Operating income was $974 million in Q2 2026, then the company reported an operating loss of $567 million in Q3, which the filing ties mainly to compute under short-term commitments. Adjusted operating income stayed positive in both quarters.
Download CSV (CC BY 4.0)
The nine-month net loss is the biggest number in the filing and the least operational. Of the $104.0 billion, $94.3 billion is a change in the fair value of convertible notes, derivatives and SAFEs, driven by a higher enterprise valuation. That is 91% of the loss, and the remainder is about $9.7 billion by our arithmetic. For comparison, the net loss for 2025 was $42.0 billion.
Claude revenue mix, customers and Claude Code
Claude's billing is mostly metered: usage fees made up 79% of the nine months to September 30, 2026 and subscriptions the other 21%. The five largest customers accounted for about 25% of nine-month sales.
The S-1/A counts customers rather than people. As of September 30, 2026 it reports:
- More than 1,900 customers with $1 million or more of trailing twelve month revenue.
- About 250 customers with $10 million or more.
- Customers with $100,000 or more grew from more than 950 to more than 11,900.
Claude Code shows up in the S-1 as a spending multiplier. In the third quarter, Claude Enterprise users who did not use it spent less than $75 a month on average, while the top 5% who did spent about 20 times that amount.
Anthropic said on February 12, 2026 that Claude Code's run-rate revenue was over $2.5 billion and that its weekly active users had doubled since January 1. It gave no absolute count, and the S-1/A reports no Claude Code user total either.
Anthropic valuation, funding and IPO status
Anthropic's post-money valuation was $965 billion after the Series H on May 28, 2026.
That is 5.3 times the $183 billion of Series F in September 2025 and 2.5 times the $380 billion of Series G in February 2026. The Series H raised $65 billion, a number that has nothing to do with the $65 billion July run rate.
Anthropic is not listed yet, but it is no longer private in the filing sense. The public S-1 was filed October 26, 2026 and amended November 2. It applies to list Class A common stock on the Nasdaq Global Select Market under the symbol ANT, and it states that there has been no public market for the shares before the offering.
How many people use Claude?
Anthropic has published no monthly user total, and its S-1 reports customers and dollars rather than people.
The measured figures are Similarweb’s 103.9 million Claude app MAU in May 2026 and 86 million claude.ai unique visitors in March to May 2026. A full-text search of the S-1/A on November 5, 2026 for "monthly active", "weekly active" and "active users" finds no matches.
Claude active users, like every other Claude user figure, count a different surface, period and method. ClaudeBot requests in a server log are crawler traffic, not human visits, as the AI crawler statistics explain.
Download CSV (CC BY 4.0)
The sources behind the middle rows are TechCrunch's March 2026 report, Similarweb's Claude and ChatGPT analysis and Similarweb's investor analysis. The 18.9 million headline sits on a page updated January 16, 2026 that names no measure, and its own website-visitor table ends at 16 million unique visitors in January 2025.
Why the 312.4 million model fails a cross-check
The largest figure is a vendor model that exceeds the sum of the measured audiences.
A user cross-check takes three steps:
- Add the measured audiences: 86 million web unique visitors plus 103.9 million app monthly active users is 189.9 million.
- Compare the model: the vendor's 312.4 million monthly users is 1.6 times that sum.
- List what the sum leaves out: Similarweb's figures omit Claude Code and the API, and the vendor says it combined 19 sources, including client campaign analytics, in a weighted model.
Treat the result as a signal only; it doesn't bound anything. The 86 million is a three-month reach, the 103.9 million is a single-month app figure, both are panel estimates, and the vendor's figure is for October, up to five months after the others.
So if you need a user number for a deck, don't quote the model. A gap of 1.6 times is large enough to ask the vendor what it counts, and nothing in the S-1/A confirms it.
Claude market share: what each measure counts
Across six audience, spend and adoption measures, Claude market share runs from 9.6% to 60.5%, because each measure uses a different denominator.
The 6.3x spread between the smallest and largest is a denominator gap, since the six measures never describe the same quantity.
Ramp publishes two separate measures. Its AI Index counts 43.8% of US businesses on Ramp with Anthropic spending in September 2026, an adoption share over all businesses on Ramp that includes subscriptions. It also reports 60.5% of AI token spending in the week ending October 23, a share that sums to 100% across vendors and counts tokens only.
You cannot subtract one from the other, because the denominators and windows differ. Quote the 43.8% as a share of US businesses on Ramp, never of all US businesses.
- Menlo Ventures, an Anthropic investor, reported that Anthropic's share of enterprise API spend increased from 40% to 55%, a 15 point move, in its 2026 report. It rests on a survey of nearly 400 US enterprise decision-makers plus a bottoms-up model, and the 40% was the figure in Menlo's 2025 report.
- Sensor Tower put Claude's US share at nearly 14% in May 2026, up from 4.4%.
Referral share is another ruler again: it counts the clicks AI answers send to websites, not the people who use Claude. Those numbers sit in ai referral traffic statistics, and the wider field of assistants in the best AI search engines.
Claude vs ChatGPT: the gap depends on the window
ChatGPT's lead over Claude depends on the window and the audience unit.
By Similarweb's count it is 12.6 times on 12-month average unique visitors (467 million against 37 million), about 5 times on unique visitors in May 2026, and 10.8 times on app monthly active users in May 2026 (1.12 billion against 103.9 million).
On visits the 12-month lead is more than 17 times, 5.7 billion against 331 million, but visits count sessions.
Download CSV (CC BY 4.0)
The window matters because Claude grew inside it: its monthly visits went from 100 million in May 2025 to 953 million in May 2026.
The overlap shifted too. The share of Claude users who also used ChatGPT in the same month fell from 78% to 61%, while the share of ChatGPT users who also used Claude rose from 2% to 14%. For the ChatGPT side of this comparison, see our chatgpt statistics.
How many businesses use Claude?
The S-1/A and Ramp answer this at different scales.
Anthropic's own filing counts customers that paid for the service: more than 1,900 with $1 million or more of trailing twelve month revenue, against over 1,000 customers spending over $1 million on an annualized basis in April 2026, up from over 500 in February.
The bases differ (trailing twelve months versus annualized spend), so the two counts do not extend one series.
Ramp's September 2026 adoption figure puts 43.8% of US businesses on Ramp paying for Anthropic against 40.6% for OpenAI. Menlo's enterprise API spend figure is a third measure, and none of the three tells you how many people at those businesses use Claude.
Is Claude losing users?
Claude is losing US mobile daily users but not web visits: Apptopia's US daily user share fell from 16.6% in July to 14.4% in August 2026, while Similarweb's visits held near 950 million a month.
So "Claude is not losing users" holds for the web only.
Three signals pull in different directions:
- Web visits plateaued at a high level. claude.ai visits went from 203 million in February 2026 to 824 million in April and then 950 million to 969 million a month, per Similarweb.
- Six-month retention trails ChatGPT but improves. ChatGPT's six-month retention is 59% against Claude's 45%, yet cohorts with a complete six-month history rise from 25.8% for January 2025 sign-ups to 44.8% for February 2026.
- US mobile slipped. Per Apptopia, Claude's daily users fell 7.0% in August 2026, 12.4% below the June peak, after its share of US daily active users had risen from 1.3% in December to 16.6% in July.
The two measurers use different units: Similarweb counts visits across the web, Apptopia counts daily users of US chatbot apps. Both can be true at once, and a plateau after a spring surge can coexist with a falling mobile share. Retention is the number to watch, because it separates users who stay from visits that spike.
Who uses Claude, and for what?
Coding is Claude's largest single use: 35% of sampled Claude.ai conversations in February 2026.
Anthropic's Economic Index puts them in Computer and Mathematical tasks, which Anthropic calls coding. The sample is 1 million conversations from Claude.ai and the first-party API, from the March 2026 report.
Download CSV (CC BY 4.0)
Why the Economic Index shares move
A share of conversations is not a share of people, and according to Anthropic's own explanations the drop in concentration has three non-user causes:
- Coding moved to the API. The decline partly reflects coding tasks migrating from Claude.ai to the first-party API, where Claude Code has grown to represent a large share of sampled traffic.
- Winter break. Some of the coursework drop is explained by academic calendars in countries where students were on winter break during the sample.
- A sign-up wave. Increasing sign-ups beginning around February brought more casual AI users.
A shift in this table becomes a hypothesis about users only after those mechanisms are ruled out. Each row is a sample of conversations, not a headcount, so read it as mix rather than audience. For measuring how AI engines send people to a site, the method belongs to AI search analytics.
Claude statistics that don't hold up
Eight widely repeated Claude figures need a correction before you quote them: two are stale, two are unconfirmed, two mislabel their measure, one is wrong and one overstates.
Download CSV (CC BY 4.0)
The count line: stale 2, unconfirmed 2, mislabeled 2, wrong 1, overstated 1.
Which Claude number to cite
Read each row as: to make this claim, cite this figure, from this measurer, and avoid the wording in the last column.
For example, a slide that says "Anthropic made $65 billion" should say instead that the run rate was $65 billion at the end of July, per CNBC's anonymous sources. You can then defend it when someone asks.
Download CSV (CC BY 4.0)
How these numbers were checked
Each figure was traced to the page that first published it, and the quoted sentence was compared with that page.
Derived numbers show their arithmetic in the sentence, and the nine-month 2026 figures are unaudited. The S-1/A user-count search, the 28-day run-rate definition and the operating figures were read in the filing itself rather than in a secondary report.
Name the surface and the window every time and a Claude number becomes quotable.
Revenue and profit are filed numbers with definitions, which is how every earlier run-rate, leak and "first profitable quarter" claim can now be tested, while the number of people who use Claude is still an outside estimate.
Before you publish, replace each bare Claude figure in your draft with a row from the table above. For grading the evidence behind GEO figures see generative engine optimization statistics, and for the wider shift see AI SEO trends.
Frequently asked questions
How much revenue does Claude make?
Anthropic's S-1 amendment of November 2 reports revenue of $17.0 billion for the three months ended September 30, 2026, about $5.7 billion a month.
It also reports $33.2 billion for the first nine months, 7.2 times all of 2025's $4.6 billion. Run rates such as the $47 billion of May describe pace, a different measure.
Is Anthropic profitable?
On a GAAP operating basis, only once: operating income was $974 million in the second quarter of 2026 and an operating loss of $567 million in the third.
Adjusted operating income stayed positive at $2.5 billion in Q3. The nine-month net loss was $104.0 billion, 91% of it a non-cash fair-value change.
Does Anthropic publish how many people use Claude?
No. The S-1 amendment reports customers and dollars, including more than 1,900 customers at $1 million or more of trailing twelve month revenue, and no monthly, weekly or daily user total. The closest measured figures are Similarweb's 103.9 million Claude app monthly active users in May 2026 and 86 million claude.ai unique visitors in March to May.
Is Claude bigger than ChatGPT?
Not on audience. By Similarweb's count ChatGPT leads Claude by 12.6 times on 12-month average unique visitors, about 5 times on May 2026 unique visitors and 10.8 times on app users.
On business spend, Anthropic held 60.5% of AI token spending among businesses on Ramp in the week ending October 23, 2026.
Is Anthropic public, and what is the Claude stock symbol?
Not yet. Its public S-1 was filed October 26, 2026 and amended November 2, applying to list Class A common stock on the Nasdaq Global Select Market under the symbol ANT.
The filing states that there has been no public market for the shares before the offering, so no stock trades under that symbol today.
Which Claude number can I quote?
Quote each figure with its measure and window.
For example, cite $17.0 billion of Q3 2026 revenue from the S-1/A, a run rate over $47 billion for May 2026 from Anthropic, and 103.9 million app monthly active users for May 2026 from Similarweb. There is no total user count to quote, so use the cite table above.
Cite this page
Aktaş, F. (2026, November 5). Claude AI Statistics 2026: Revenue, Users and Which to Cite. Mission Growth. https://missiongrowth.io/blog/claude-statistics
Download the data: 9 tables as CSV (CC BY 4.0)
- Anthropic recognized revenue by period, 2025 to 2026
- Anthropic run-rate revenue timeline, 2025 to 2026
- Anthropic profit and loss measures, 2026
- Claude user estimates by measure and source, 2026
- Claude market share by measure and denominator, 2026
- ChatGPT lead over Claude by surface and window, 2025 to 2026
- Claude usage mix from the Anthropic Economic Index, 2025 to 2026
- Claude figures that need correction, 2026
- Which Claude number to cite for which claim, 2026
Figures we made for this post are free to reuse under CC BY 4.0 with credit to Mission Growth.