Free Made ChatGPT Famous. Paying Customers Are Choosing Claude.
There's a habit in technology coverage of measuring winners by raw user counts. It's a reasonable proxy, most of the time. But paid subscribers are different. They're the signal underneath the noise, the people who've tried the options, decided one is worth money, and keep deciding that every month.
By that measure, the chatbot market is sending a signal that most coverage has treated as background noise. Claude's paying user base and revenue grew approximately 75 percent between January and May 2026, according to Indagari, which analyzes anonymized credit card transaction data from roughly 28 million U.S. consumers. ChatGPT still has more paid subscribers overall. That's not in dispute. But 75 percent growth, sustained across a five-month window that included some serious Anthropic turbulence, is worth examining carefully.
The learning platform data adds another dimension. DataCamp, which serves around 20 million learners worldwide, reported this month that demand for Claude courses had increased 18 times over in the last 30 days. Among self-directed learners spending their own time and money to develop AI skills, Claude courses now outpace ChatGPT courses three to one. These aren't enterprise seat purchases driven by procurement. These are individuals who've used both tools and decided one is worth mastering.
The web visit numbers put both data points in context. ChatGPT still leads global web traffic with roughly 55 percent of chatbot visits, ahead of Google Gemini at 27 percent and Claude at about 8 percent. But Claude's visit count grew 306 percent in a single quarter, from 203 million visits in January to 824 million in April. When you combine that growth velocity with disproportionate gains among paying users, two separate signals are pointing at the same conclusion.
What's driving the shift? Several theories are in circulation, and none fully explains the gap on its own.
Writing quality for professional work is the most commonly cited factor. Knowledge workers who rely on AI for long-form output, analysts, researchers, lawyers, and consultants, report that Claude handles sustained, structured writing without the over-hedging and confident incorrectness that frustrate ChatGPT users in extended work. Whether the gap is as large as the anecdotal reports suggest is genuinely hard to measure. But users are measuring it themselves, with their wallets.
The more interesting factor might be the March 2026 moment. Anthropic declined to allow Claude's models for use in mass surveillance or autonomous weapons systems. The decision generated coverage and some controversy. But a notable portion of Claude's subscriber growth continued through and after that announcement. For some users in research, policy, and regulated industries, the vendor's stated values appear to be part of the product. That Anthropic's government access restrictions made headlines again in June suggests this isn't a settled question, though it hasn't reversed the growth trend.
There's also a structural story worth watching. ChatGPT courses on DataCamp remain more popular for corporate training programs, where organizations buy seats at scale based on name recognition and procurement inertia. Claude's edge is concentrated among individual learners. The enterprise market and the individual market are related but distinct, and their preferences often diverge for years before one leads the other. Microsoft locked in enterprise email with Outlook and only later watched consumers drift to Gmail. The analogy isn't perfect, but the pattern of enterprise incumbency and individual-market erosion isn't rare.
None of this destabilizes OpenAI's position in the near term. GPT-5.6, despite a government-requested delay, will reach a large and already-converted user base. The API integrations, the partnerships, the embedded tooling in IDEs and productivity software compound in ways that take years to erode. Google's Gemini 2.5 Pro, meanwhile, leads benchmarks in several categories and has its own distribution advantages. ChatGPT is not losing the chatbot market.
But here's the uncomfortable question the usage data raises: how much of ChatGPT's paid subscriber base is there because users actively prefer it, and how much is there because switching felt like too much work? That distinction matters a lot less when you're growing. It matters considerably more when a competitor is demonstrating 75 percent revenue growth in five months and outpacing you three to one among people actively choosing who to learn from.
Free made ChatGPT the default for an enormous slice of the internet. Defaults are powerful until they're not. The history of technology is full of defaults that held right up until they didn't. The credit card data and the learning platform trends suggest that the subset of users paying close attention have already started making a different choice. Whether that moves upstream into broader behavior is the question. For now, it's a signal worth watching.
Want a weekly digest of what's shifting in the chatbot market? Subscribe to About.chat Weekly — free, no spam.