AGENTIC COMMONSAI industry briefings

繁中EN

TagAnthropicPublished

All English articles

Paid Consumers Flock to Claude as ChatGPT's Moat Cracks

On this page6 sections
  1. Three Sources, One Direction
  2. What The Payment Records Show
  3. The Signal Learners Send First
  4. Measured By Scale, ChatGPT Stays In Front
  5. A Fight With Government Meets The IPO Clock
  6. References

Summary: TechCrunch compiled three independent datasets on June 25 showing accelerating paid consumer momentum for Claude. Indagari’s card data found paying Claude consumers and revenue up about 75% since January 2026, while DataCamp saw Claude become its most-searched keyword. Sensor Tower still shows ChatGPT far ahead in downloads, active users, brand recognition, and total paying users.

For years Claude has worn a single label: the model built for engineers and enterprises. Code gets written in Claude Code, knowledge work runs through corporate deployments, and the consumer side of the market was simply handed to ChatGPT. Three separate data sets, compiled by TechCrunch on June 25, put a question mark against that division of territory. Paying individual consumers are moving toward Claude at a quickening pace, and a market that once leaned entirely one way is loosening at the edges. Calling it a reversal would be premature — yet the direction of paid consumer AI has, for the first time, become genuinely contested.

Three Sources, One Direction

None of the three sources knows about the others. Indagari, a consumer finance data firm, breaks down credit-card transactions; DataCamp, an online learning platform, publishes search and course statistics; and Sensor Tower issues the State of AI 2026 report. Each one, taken alone, can be criticized for bias. Placed side by side, they get hard to argue with — the paid momentum Claude has built among consumers no longer fits inside the developer world where it began.

What The Payment Records Show

Indagari works from billions of anonymized transactions covering roughly 28 million U.S. consumers, and it logs subscription and API token spending week by week, from 2025 through May 10, 2026. The finding fits on one line: since January 2026, the number of paying Claude consumers and the revenue they generate are up about 75%. A caveat belongs right beside that figure — data of this kind describes a trend, not an absolute size. Claude’s base sits far below ChatGPT’s, and nothing in these numbers reveals how many subscribers either company truly has. Even so, the growth curve has visibly steepened, and card records carry something surveys never can: a log of what money was actually spent on, rather than what people claim to use. A 75% gain measured at the checkout is not something spin can explain away.

The Signal Learners Send First

DataCamp serves roughly 20 million users, and activity inside its platform speaks loudly. The most-searched keyword on the site is now “Claude,” which has pushed “AI” out of the top spot — the name it displaced is not “ChatGPT,” not “Python,” but the proper name of a single AI product. Among self-directed learners, Claude courses draw triple the demand of ChatGPT courses, and over the last 30 days that demand multiplied 18x. Corporate training flips the picture: ChatGPT stays on top there, a tidy portrait of how differently the two audiences are assembled. Claude wins individual professionals first; organizational purchasing remains with ChatGPT. Searches run ahead of the market — people start paying to learn a tool shortly before they commit their careers to it.

Measured By Scale, ChatGPT Stays In Front

State of AI 2026 from Sensor Tower is the counterweight. Claude is climbing hard on every platform, yet the distance separating it from ChatGPT is still enormous — downloads, monthly active users, and brand recognition among people who have never written code all say so. ChatGPT’s own growth rate looks modest, chiefly because its base is already so large, and it keeps hold of far more paying users. Restated: Claude is winning pace right now, not size. That distinction offers OpenAI no comfort, and it hands Anthropic no victory either. At this scale, subscriber loss runs in both directions. ChatGPT has to answer whether the $20-a-month wall survives as free alternatives keep improving; Claude has to answer whether subscribers stay once the news cycle cools. Both companies are headed for public listings, and investors will read both questions closely.

A Fight With Government Meets The IPO Clock

The calendar gives the growth story its shape. Claude’s consumer-side climb began with a distinct March spike, right after Anthropic refused to let the U.S. government apply its models to mass surveillance and autonomous weapons. In June, after Washington barred Anthropic’s Mythos 5 and Fable 5 from non-American use, the company withdrew both models from sale altogether. TechCrunch’s June 16 report offers a reading of that standoff: the clash with the administration may have ended up feeding Anthropic’s sales — to many individual subscribers, a company that will stand up to its own government looks like a company that deserves their money. The other backdrop is financial: Anthropic submitted its IPO filing on June 1, 2026, and OpenAI is traveling a similar road, so the paying-consumer growth curve will land directly in the prospectus. Of all the revenue lines an AI company can put in front of investors, consumer subscriptions carry the highest margins and the greatest visibility. Anthropic declined to comment on the coverage.

References

AGENTIC COMMONSOperated by PHLEGON LABS
ShareXEmail
Support us

Related reading

  1. Anthropic-GSA Deal: 40% Off Claude for All Government Levels

    Anthropic's GSA agreement adds a 40% OneGov discount on Claude Enterprise, Claude for Government, and the API for federal through local agencies.

    Anthropic

  2. Two-Tier Token Pricing and Effort Knobs Change How You Route Cheap Work

    Haiku 5.5 adds tiered long-context pricing and per-task effort controls, reshaping how you route high-volume agent work.

    Claude

  3. Frontier Coding Output, Now Priced Like a Mid-Tier Model

    Claude Opus 5.5 matches Fable 5.1-level work while cutting typical run costs 40% versus Opus 5, led by cheaper cache reads.

    Claude