Claude's clarity beats OpenAI's chaos

Anthropic has done more right in three months than OpenAI did in all of last year. And the crazy thing is: it wasn’t about the model, but simply about focus.

The numbers

Anthropic confirmed to TechCrunch that paid Claude subscriptions have more than doubled since the start of the year. The majority of new customers are signing up for the $20/month Pro tier.

This isn’t just self-reported data. The analytics firm Indagari analyzed credit card data from 28 million U.S. consumers for TechCrunch. The result: a record number of new customers between January and February. A record number of returning users in February. Consumers are spending money on Claude at a record pace.

Two caveats: The data does not capture all consumers and does not include enterprise customers or free-tier users. Claude still trails ChatGPT in total paid subscribers. Anthropic itself does not provide overall figures—estimates range between 18 and 30 million.

Why now?

Three things happened almost simultaneously, and none of them was a coincidence.

The Super Bowl ad. Anthropic ran several Super Bowl ads that mocked advertising in AI chats. Tagline: "Ads are coming to AI. But not to Claude." OpenAI wasn’t named—nor did it need to be. The result: Claude climbed into the top 10 of the app stores, with an 11% increase in daily active users.

The Pentagon refusal. Anthropic refused to make Claude available for autonomous weapons systems and mass surveillance of U.S. citizens. The Pentagon subsequently placed Anthropic on a risk list. Anthropic sued. That sounds like a PR disaster—but it was the opposite. Because OpenAI struck its own DoD deal at the same time. The result, according to Sensor Tower via TechCrunch: ChatGPT uninstalls rose by 295% the day after the deal was announced—compared to a normal daily rate of 9%.

The product offensive. Claude Code, Cowork, and Computer Use—three launches in quick succession. Not features for the press release, but tools that developers and teams actually use. Meta has adopted Claude Code internally to such an extent that the company is organizing an entire "AI Transformation Week" centered around Anthropic’s tools—hackathons, demos, and training sessions for all levels of the organization. CEO Mark Zuckerberg calls 2026 the year in which AI will "dramatically change the way we work." The fact that he’s relying on Claude’s tools for this—alongside his own Llama models—says a lot.

Enterprise: The Quieter Front

The consumer numbers are making headlines. The enterprise story is more important in the long term.

Anthropic is targeting $20–26 billion in revenue for 2026. According to a February 2026 a16z survey, 78% of the enterprise CIOs surveyed use OpenAI in production—and 44% use Anthropic, with the trend on the rise. OpenAI has scaled its enterprise sales team from 10 to 500 people in under two years. Anthropic is growing more slowly, but wallet share is shifting.

Let’s be honest: if nearly half of all CIOs have Anthropic in production, it’s no longer an underdog. It’s a second standard.

What this means for OpenAI

OpenAI faces three problems at once: its consumer base is eroding, its enterprise lead is shrinking, and its “side quests” (Sora, Instant Checkout) have failed. Fidji Simo, OpenAI’s CEO of Applications, internally called Anthropic’s success a “wake-up call.” The response: a return to coding tools and enterprise. More like “dare to be Anthropic,” so to speak.

The promise: OpenAI as a platform for everything. The reality: A company that has to figure out what it actually wants to be—while its competitor already knows exactly that.

The real point

Anthropic’s growth isn’t a product-feature comparison. It’s a strategy built on three pillars: ethical positioning that resonates with consumers and developers alike. Products that transform workflows rather than just delivering demos. And a focus that has never wavered.

Claude isn’t better than GPT. Claude is clearer than OpenAI. And in a market where everyone promises the same thing, clarity wins.

An LLM researched and wrote. A human read, edited, and approved it. We’re still debating which of the two did more work.