Spending data and search trends point to a quiet shift among consumers who actually open their wallets for AI, even as ChatGPT keeps a commanding lead in raw user numbers.
Anthropic has spent most of its life being cast as the AI lab for builders, the tool that coders and enterprises reach for while everyone else opens ChatGPT. Fresh spending data complicates that storyline.
Credit card transaction analysis from the firm Indagari shows paying consumers steadily moving toward Anthropic's Claude through the first half of 2026. Paid users and revenue in this segment have risen about 75% since January, climbing month after month rather than spiking once and flattening. The dataset pulls from billions of anonymized transactions tied to roughly 28 million U.S. consumers and tracks weekly payments from 2025 through May 10 of this year, covering subscriptions as well as API token purchases.
That sample will not hand anyone Anthropic's exact revenue or a precise customer count. It is wide enough to show direction. And the direction keeps pointing up.
A customer base wider than the developer crowd
For two years the working assumption was simple: Claude's audience was narrow. Software engineers who loved Claude Code and startups that wired the API into their products made up the core. The mainstream consumer, the thinking went, belonged to OpenAI.
The transaction figures push back on that. The recent climb is being driven by individuals paying out of pocket, not corporate accounts, which suggests Anthropic's reach now stretches past the technical base that first championed it.
There is a second tell worth examining beyond raw dollars.
What DataCamp's search data adds
DataCamp, an online platform that teaches AI skills to consumers and corporate employees, says it has roughly 20 million users. Interest in Claude on the site has shot up since January.
"Claude" is now the single most searched term across the platform. It has passed even the word "AI" itself, the company told TechCrunch.
The pattern splits by audience. For corporate training programs, ChatGPT courses still draw the bigger crowd. Among self-directed learners signing up on their own, demand for Claude material is running ahead of ChatGPT by three to one. DataCamp says course demand around Claude jumped 18-fold in a single 30-day stretch.
Two independent signals, one from spending and one from search behavior, are bending the same way.
The fight that may have helped
Some of the momentum traces back to a confrontation that had nothing to do with product features.
Earlier this year Anthropic refused to let its models be used by the Trump administration for mass surveillance of Americans or for autonomous weapons. The Pentagon retaliated by labeling the company a supply chain risk, and federal agencies began phasing Claude out. Anthropic sued, and that litigation is still working its way through the courts.
What happened next surprised even the people tracking it. Rather than scaring off the public, the standoff appeared to deepen consumer loyalty. Claude vaulted to the number one spot on Apple's chart of top free U.S. apps in late February, and the spending surge that followed in March did not fade afterward. The gains carried into the months that came next.
A principled stance, it turns out, doubled as marketing.
ChatGPT still owns the room
None of this means Claude has caught its rival. It has not, and the gap is large.
Market intelligence firm Sensor Tower shows Claude growing across every platform this year while sitting well behind ChatGPT on overall usage. The Indagari data tells a matching story on payments: OpenAI still counts far more paying users than Anthropic does. ChatGPT's recent growth has looked more modest by comparison, though that has less to do with weakness and more to do with the sheer size of the base it already commands. When a product serves hundreds of millions of people, big percentage jumps get harder to post.
Put plainly, Claude is gaining ground on ChatGPT in consumer dollars and brand awareness while remaining the clear number two.
An IPO race raises the stakes
The timing of all this matters because both labs are walking toward Wall Street.
Anthropic confidentially filed a draft S-1 with the Securities and Exchange Commission on June 1, getting out ahead of OpenAI's expected filing. The move landed days after a $65 billion Series H round pushed the company's valuation to $965 billion, edging past OpenAI's $852 billion. Anthropic has said its annualized revenue run-rate crossed $47 billion, up from around $9 billion at the close of 2025.
OpenAI is preparing its own offering, and SpaceX has already filed publicly, setting up one of the busiest stretches of mega-listings the tech sector has seen.
A public offering would force open a door that has stayed shut. Investors have spent the past year debating whether AI valuations rest on real demand or on narrative, and an S-1 carries audited financials, risk disclosures, and a breakdown of which products actually generate the money. Until those numbers surface, third-party transaction data and search trends are the closest thing to a window anyone gets.
There is one more data point that should interest the bankers. Fintech firm Ramp reported that more businesses used Anthropic than OpenAI for the first time in May, a sign the consumer shift may have a commercial twin.
The Mythos and Fable wildcard
Hanging over the optimistic read is Anthropic's latest collision with Washington.
On June 13 the U.S. government issued an export control directive barring foreign nationals from accessing the company's two most capable models, Mythos 5 and the cybersecurity-focused Fable 5 built on it. The order applied to any foreign national inside or outside the country, including Anthropic's own non-American staff.
The company could not verify the nationality of every user on every request, so it complied the only way it could. It switched both models off for everyone, Americans included.
Anthropic pushed back hard in public. It said the government's stated basis was a narrow technique for bypassing Fable 5's safeguards, one it argued could pull similar capabilities out of other widely available models that face no comparable restrictions. Recalling a model deployed to hundreds of millions of people over a finding that limited, the company said, would set a standard capable of freezing new releases across the entire frontier model industry.
Critics offered a sharper reading. One cybersecurity researcher noted on X that a company describing its product as a weapon-grade tool in launch after launch should not be shocked when regulators take the description at face value. Anthropic, in this view, wrote the legal predicate itself and stamped a brand on it.
For now the most powerful models stay dark while the directive holds, an awkward backdrop for a company courting public investors.
The open question
Every clean read of the available data shows Anthropic expanding its consumer and enterprise footprint at the same time, government friction notwithstanding. What no one outside the company can yet measure is how much the running battles with Washington will cost the business once the books are open.
Comments
Join the discussion and share your perspective.