The company that didn't exist at commercial scale 18 months ago just posted bigger revenue than OpenAI and is now eyeing a near-trillion-dollar IPO.
The Summary
- Anthropic generated $11.5-12B in Q2 revenue, a 14x jump from last year, surpassing OpenAI's quarterly revenue for the first time
- The company raised $65B in Series H funding while its Claude Opus 5 model leads current AI rankings
- Anthropic is preparing for a potential 2026 IPO at a $965B valuation, which would be the largest public offering of the year, with internal targets reaching $2T by 2028
- The company is in talks to acquire Israeli AI startup Decart for $7B, signaling aggressive expansion beyond foundation models
The Signal
Anthropic's revenue growth isn't just impressive, it's structurally different. The company went from near-zero commercial revenue to $11.5 billion in six months. That's a 1,500% increase year-over-year, but the more interesting number is how they got there. While OpenAI built its empire on ChatGPT consumer subscriptions and API access, Anthropic went straight to enterprise. Claude isn't winning on brand recognition. It's winning on contracts.
The $65 billion Series H round puts Anthropic's total raised north of $100 billion across all funding stages. That's not venture capital anymore. That's nation-state level resource allocation. The money is going into compute, talent, and something OpenAI lost two years ago: the ability to tell Fortune 500 CTOs that your model won't train on their proprietary data unless they explicitly opt in.
"Anthropic went from near-zero commercial revenue to $11.5 billion in six months."
Here's what matters for the agent economy:
- Enterprise trust compounds faster than consumer virality
- Constitutional AI (Anthropic's safety approach) is becoming table stakes for regulated industries
- Companies building agents need foundation models they can audit and constrain
The potential $7 billion Decart acquisition tells you where Anthropic thinks the next revenue layer lives. Decart specializes in real-time AI video generation and simulation. That's not about chatbots. That's about agents that can navigate visual interfaces, generate training data for robotics, and build synthetic environments for testing. Anthropic is buying the ability to move from text to action.
The IPO math is aggressive but not absurd. A $965 billion debut valuation would value Anthropic at roughly 80x trailing quarterly revenue. Nvidia trades at 35x. But Nvidia sells shovels. Anthropic is selling the gold rush itself. If the company hits its internal targets and reaches $2 trillion by 2028, it would need to sustain roughly 100% year-over-year growth. Impossible? Two years ago, posting $12 billion in a quarter would have seemed impossible.
The Implication
Watch what Anthropic does with the Decart acquisition. If it closes, you're looking at the first major foundation model company to vertically integrate into specific agent capabilities rather than staying horizontal. That changes the build-versus-buy calculus for every company trying to deploy autonomous systems. If Anthropic can offer not just the language model but the vision, simulation, and action layers in one package, OpenAI's API-first strategy starts looking brittle.
For builders: enterprise revenue beats consumer buzz. For investors: the AI market isn't consolidating, it's stratifying. There will be consumer AI companies and there will be enterprise AI companies, and they'll barely compete. Anthropic just claimed the enterprise crown while everyone was watching ChatGPT's download numbers.