For most of the past four years, the story of generative AI has been told through one lens: OpenAI’s. GPT-3, GPT-4, ChatGPT, the $80-billion valuation, the Sam Altman drama, the partnership with Microsoft — the San Francisco lab has dominated headlines and mindshare in a way that no other AI company has come close to matching. That narrative is now cracking.
In a shift that would have seemed implausible even twelve months ago, Anthropic has surpassed OpenAI in annual revenue, posting a $47 billion annualized run rate against OpenAI’s self-reported figure of $25 to $33 billion. The gap is not narrow. Anthropic is running nearly double the revenue of the company that invented the modern AI era — and it did so without a single viral consumer product moment, without billions in Microsoft integration, and without a single day of being publicly traded.
Claude Code Changed Everything
Ask Anthropic executives what drove the inflection and the answer is consistent: Claude Code. The company’s agentic coding assistant — launched in earnest in late 2025 — became the single most important product in the company’s history almost immediately. Enterprise developers adopted it for everything from automated code review to full-stack feature generation, and unlike a ChatGPT subscription that might sit mostly idle, Claude Code was billing heavily every single day.
The numbers tell the story bluntly. Just twelve months ago, Anthropic’s annualized revenue stood at roughly $1 billion. By the end of 2025, it had climbed to $9 billion. By April 2026, it had surged to $30 billion. By May 2026, it had rocketed past $47 billion. That trajectory — a 47x increase in revenue in under a year — is the kind of growth curve that gets written about in business school textbooks. The underlying driver is enterprise software adoption mechanics: once a developer team integrates Claude Code into their workflow, usage compounds. More projects, more tokens, more billing — every quarter.
Two Very Different Bets on How AI Makes Money
Anthropic and OpenAI have made structurally different bets on how AI monetization works, and those bets are now resolving.
OpenAI bet on consumers. ChatGPT’s subscription model, which charges individuals $20 to $200 per month, generates enormous brand awareness and cultural relevance. The product is genuinely popular — ChatGPT remains among the most visited websites on the internet. But consumer subscriptions have natural ceilings. A doctor pays $200 a month whether she uses the product twice or two thousand times. The revenue is capped at the subscription price.
Anthropic bet on enterprise. Claude’s API pricing, enterprise contracts, and usage-based billing mean that heavy users generate heavy revenue. A bank processing millions of documents through Claude’s API, or a software company running Claude Code across a thousand-engineer team, generates a fundamentally different revenue profile than any number of individual subscribers. The enterprise model is stickier, more predictable, and — critically — not capped. As enterprise customers grow, their AI spend grows with them.
The result is that Anthropic’s revenue has scaled in direct proportion to how deeply enterprises have embedded Claude into their operations. And over the past year, that embedding has been aggressive.
The Profitability Question
One of the most striking aspects of Anthropic’s position is that it achieved profitability in Q4 2025 — an accomplishment OpenAI has not yet reached despite processing vastly more consumer transactions. This matters for more than bragging rights. A profitable AI lab is an AI lab that controls its own destiny. It can invest in safety research, hire talent, and build products without being perpetually dependent on the next fundraising round to stay solvent.
Anthropic’s $65 billion Series H round, co-led by Altimeter Capital, Dragoneer, Greenoaks, and Sequoia Capital, pushed the company’s valuation to $965 billion — formally surpassing OpenAI’s last reported valuation of $852 billion and making Anthropic the most valuable privately held AI company in the world. The round was not a lifeline; it was an offensive move. Anthropic raised that capital from a position of strength, at a moment when its revenue trajectory gave it every negotiating advantage.
The company also filed a confidential draft S-1 registration statement with the SEC on June 1, 2026, with a potential public listing expected as early as October 2026. Goldman Sachs, JPMorgan, and Morgan Stanley are leading an offering expected to raise more than $60 billion. That would make the Anthropic IPO one of the largest technology offerings in history.
What This Means for the AI Industry
The revenue reversal carries implications that extend well beyond the competitive dynamics between two San Francisco AI labs.
First, it validates the enterprise-first AI business model at a scale that previously existed only in theory. For years, AI optimists argued that the real money in generative AI would come from B2B applications — not consumer subscriptions — because enterprises could convert AI productivity gains into dollars in ways that individuals could not. Anthropic’s $47 billion run rate is the empirical proof of that thesis.
Second, it puts enormous pressure on OpenAI’s IPO strategy. OpenAI is also preparing to go public, and its S-1 will inevitably be compared against Anthropic’s. If Anthropic can demonstrate profitability, faster revenue growth, and higher enterprise retention, OpenAI will face difficult questions from institutional investors about its path to sustainable economics. The companies are no longer competing only on model quality benchmarks; they are competing on financial fundamentals.
Third, it signals that the AI market is large enough for multiple winners — but the nature of those wins may look very different. OpenAI may remain the dominant consumer AI brand for years. Anthropic may come to own the enterprise infrastructure layer. Google, with Gemini deeply embedded in Workspace and Cloud, occupies a third distinct position. The AI market is not consolidating into a single winner; it is stratifying.
The Safety Angle Nobody Is Talking About
There is an aspect of Anthropic’s revenue milestone that has received almost no coverage: what it means for the company’s safety research mission.
Anthropic was founded explicitly on the premise that safe AI development requires being at the frontier — that you cannot influence AI safety from the outside. For that to work, the company needed to remain financially competitive with OpenAI and Google indefinitely. For most of its existence, that required large external injections of capital and a certain amount of faith that the bet would eventually pay off.
Now, for the first time, Anthropic’s safety mission and its commercial reality are aligned. The company that built Claude with constitutional AI, that publishes interpretability research, that has arguably invested more in AI alignment than any other lab, is also the most financially successful AI company in the world right now. The argument that safety and commercial success are in fundamental tension — an argument that has animated much of the AI industry debate for years — has taken a significant empirical hit.
What Comes Next
The next six months will be decisive for both companies. Anthropic is expected to launch new model generations and expand Claude Code’s capabilities into domains beyond software development. OpenAI is pushing hard to broaden GPT-5.6’s availability beyond its current government-vetted partners while also rolling out GPT-Live, its new full-duplex voice model.
Both labs are investing heavily in agentic AI — systems that can execute multi-step tasks autonomously without constant human oversight. If the last year was defined by coding assistants, the next year may be defined by AI agents that handle entire workflows. Whichever company builds the most trusted, most capable, and most reliable agentic platform will likely extend or reverse the current revenue gap.
For now, the headline is clear: Anthropic has overtaken OpenAI in revenue, is profitable, and is approaching a trillion-dollar IPO. The company that critics once dismissed as “the safety lab that couldn’t compete” has built, quietly and methodically, the most commercially successful AI business in the world.
The AI race has a new leader. The question is how long they can hold the position.
Sources: Fortune (July 2, 2026), CNBC, Neowin, Schemaninja, Anthropic official press releases
