Here's a number that stopped me cold: Anthropic's annualized revenue hit $19 billion. That's 1,167% year-over-year growth. For context, the company earned its first dollar in revenue roughly three years ago. It has since grown over 10x every single year.
The source isn't a press release or a pitch deck. It's anonymized credit card transaction data from 28 million U.S. consumers, reported by TechCrunch on March 28. Claude paid subscriptions have more than doubled in 2026, and the trend was still accelerating through early March. People aren't just trying Claude. They're pulling out their wallets.
What makes this particularly interesting isn't the raw growth. It's the combination of factors that created it, and whether they're repeatable.
The Pentagon Effect
In late January, Anthropic and the U.S. Department of Defense publicly clashed over the use of Claude's AI models. Anthropic drew a hard line: no lethal autonomous operations, no mass surveillance of American citizens. The DoD wasn't thrilled. The story made national news.
What happened next was unexpected. TechCrunch reported on March 6 that previously churned users returned to Claude in record numbers. Between January and February, the return rate hit a historical high. Consumers didn't just notice the Pentagon dispute. They rewarded Anthropic for it.
This is the "Patagonia effect" applied to AI. When a company takes a principled stand that costs it something, a segment of consumers responds by opening their wallets. Anthropic turned a government contract dispute into a marketing event, and it wasn't even intentional.
The Super Bowl Gambit
Anthropic ran Super Bowl ads that directly targeted OpenAI. The message: ChatGPT is showing you ads. Claude never will.
The timing was strategic. OpenAI had announced ad integration for ChatGPT's free tier, creating immediate anxiety among users who'd grown comfortable with an ad-free experience. Anthropic's ads essentially said, "If you don't like where ChatGPT is heading, we're the alternative."
The Super Bowl spots went viral. They sparked a wave of ChatGPT-to-Claude migration conversations across developer communities, Reddit, and X (formerly Twitter). Whether those conversations converted to sustained subscribers is the real question, but the January-February credit card data suggests many of them did.
Post not found or has been removed.
Claude Code: The Durable Growth Engine
The Pentagon dispute and Super Bowl ads are one-time events. They spiked awareness but won't keep subscribers paying month after month. The product that actually retains paying users is Claude Code.
Claude Code launched publicly in May 2025. Less than a year later, it has run-rate revenue above $2.5 billion. Business subscriptions quadrupled since January 2026. Enterprise usage represents over half of Claude Code revenue. According to SaaStr, no B2B software product in history has grown this fast.
The March feature blitz amplified this. Anthropic shipped 14+ features in March alone: Computer Use for desktop automation, auto mode for permission-free coding, hooks for custom automation pipelines, enhanced agents for parallel tasks. Each feature makes Claude Code stickier for its existing users and more attractive to new ones.
Claude Code usage grew 300% since the Claude 4 model launch. Run-rate revenue is up 5.5x. Anthropic also shipped an enterprise analytics dashboard for tracking spend and code acceptance rates, signaling that large organizations are Claude Code's highest-value customer segment.
Post not found or has been removed.
The Traffic Numbers
Semrush data fills in the picture. Claude.ai hit 220 million monthly visits in January 2026, then jumped 30.9% to 288 million in February. Meanwhile, the Model Context Protocol SDK hit 97 million monthly downloads in March, up from 2 million at launch in November 2025. That's 4,750% growth in 16 months. The developer ecosystem is building momentum that feeds back into consumer adoption.
Most new subscribers are on the Pro plan at $20 per month. This is the key signal that Claude is expanding beyond its power-user base into the general consumer market. Higher tiers exist ($100 and $200), but the mass-market growth is happening at the entry level.
5 Outages in March
Growth this fast breaks things. The New Stack documented that Anthropic suffered 5 service outages in March while shipping 14+ features. A major incident on March 2 took down claude.ai for hours. Errors spiked again on March 11 and March 25. On March 27, Opus 4.6 and Sonnet 4.6 models experienced concentrated failures.
The root cause: networking performance degradation between serving stack components. Anthropic migrated workloads to healthy infrastructure to recover. The pattern is familiar: growth is outrunning infrastructure scaling, and the engineering team is fighting to keep up.
For enterprise customers, reliability is a hard requirement. Five outages in one month is a credibility problem, regardless of how impressive the feature velocity is. Anthropic needs to solve this if it wants to keep converting enterprise trials into long-term contracts.
Post not found or has been removed.
ChatGPT vs. Claude: A Different Kind of Race
ChatGPT still has far more total users. That's not in dispute. But the credit card data reveals something the user-count comparison misses: Claude is winning on conversion rate.
ChatGPT's massive free tier means millions of users who never pay. Claude's free tier is more limited, pushing higher proportions of engaged users toward paid plans. Anthropic isn't trying to win on total users. It's trying to win on revenue per user.
The ad model divergence amplifies this. As ChatGPT integrates advertising into its free experience, users who specifically value an ad-free product have a clear alternative. Anthropic is betting that enough users will pay $20/month to avoid ads, and the credit card data says that bet is paying off.
The question isn't whether Claude can overtake ChatGPT in total users. It probably can't, at least not soon. The question is whether Claude can build a more profitable business with a smaller but more paying user base. The $19 billion ARR number suggests the answer might be yes.
Credit card transactions are the purest signal in tech. They show what people actually value, not what they click on. In 2026, an unprecedented number of people decided Claude was worth paying for.
Full Korean analysis on spoonai.me.
Sources:
Top comments (0)