Claude Don't Lie: Anthropic's Revenue Triples in Four Months, Makes OpenAI's 2030 Breakeven Look Like a Bug

Anthropic's revenue trajectory keeps embarrassing its own CEO.

The company dropped some numbers this week that would make even the most bullish AI maxis do a double-take. Annualized run rate crossed $30 billion as of early April 2026, up from approximately $9 billion at the end of 2025. For context, Anthropic started 2025 at $1 billion, hit $4.5 billion by mid-year, stumbled past $9 billion in December, reached $14 billion in February during its Series G, and then just... kept going. At this point, we're basically watching a different kind of diamond hands play out.

CEO Dario Amodei has publicly admitted he's been wrong about projections "every time," claiming he stays "very conservative." Apparently, conservative means off by a factor of three in four months. Either Dario has the worst crystal ball in Silicon Valley, or these numbers are actually designed to make investors feel like they underspent on the Series G. Either way, we're not complaining.

What's driving it?

Claude Code, the company's agentic coding platform, is doing heavy lifting. It generated over $2.5 billion in run-rate revenue as of February, with weekly active users doubling since January 1. That's not growth—that's a supply shock. Developers apparently decided that having an AI do their busywork was worth $2.5 billion annually. Smart money, honestly.

On the enterprise side, the numbers are equally absurd. Customers spending more than $1 million annually now exceed 1,000—doubling from 500 in under two months. These aren't companies dabbling in AI for the vibes. They're deploying it as core infrastructure for legal, finance, consulting, and communications. Knowledge worker productivity apparently has a very real price tag. Spoiler: it's $1 million per year, minimum.

Claude's API market share expanded from 12% in 2023 to 32% by mid-2025, overtaking OpenAI for enterprise language model dominance. OpenAI went from the cool kid at the party to that guy who keeps texting unopened. The shift happened faster than most predicted, and slower than Anthropic will admit they expected.

Anthropic also announced a compute deal with Google and Broadcom, securing access to approximately 3.5 gigawatts of TPU capacity starting in 2027. This extends the $50 billion US AI infrastructure commitment announced in November 2025. The company already runs workloads across AWS Trainium, Google TPUs, and Nvidia GPUs, matching workloads to the right silicon. Essentially, they're building a diversified mining operation—but for AI tokens instead of proof