In 2022, OpenAI made 20 times more revenue than Anthropic. By early 2026, Anthropic's annualized revenue run rate reportedly surpassed OpenAI's at $30 billion. That is not a gradual closing of the gap. That is the fastest revenue growth in enterprise software history.
This is not a prediction piece. It is an analysis of where both companies stand right now, based on publicly reported financial data, model benchmarks, and market positioning. For a deeper look at their flagship models, see our Claude vs ChatGPT comparison.
Who This Is For
- Investors and analysts tracking the AI infrastructure market
- Enterprise buyers deciding between OpenAI and Anthropic APIs
- Anyone following the most consequential tech rivalry since Apple vs Google
Quick Verdict
OpenAI has the brand recognition and consumer dominance. Anthropic has the growth rate, the developer loyalty, and the better gross margins. If the race is measured by cultural impact, OpenAI leads. If measured by business fundamentals and trajectory, Anthropic is pulling ahead.
By 2028, the most likely outcome is not one winner. It is a duopoly — similar to AWS vs Azure in cloud computing. Both will be massive. The question is which one becomes the default for developers and enterprises.
The Revenue Numbers
This is where the story gets surprising. According to Reuters, at the start of 2025, OpenAI led Anthropic in annualized revenue by $6 billion vs. $1 billion. By late 2025, Anthropic had reached $9 billion. By February 2026, that doubled to $14 billion. By April 2026, reports placed Anthropic's run rate at $19-30 billion.
The growth rate: Anthropic has been growing at roughly 10x per year. OpenAI has been growing at 3.4x per year. At those rates, Anthropic's revenue overtakes OpenAI's by mid-2026 — and the gap widens from there.
Forbes reports that Anthropic's gross margins are projected to reach 50% in 2025 and 77% by 2028, compared to OpenAI's approximately 33%. Higher margins mean Anthropic generates more profit per dollar of revenue — a critical advantage as both companies scale.
Why the Models Matter
Revenue follows model quality. Here is where each company's latest offerings stand:
Coding and engineering: Claude Opus 4.6 leads SWE-bench Verified at 78.7%, edging out GPT-5.4 at 76.9%. But GPT-5.4 leads on HumanEval (93.1% vs 90.4%). The pattern: Claude wins at real-world software engineering tasks. GPT wins at synthetic coding benchmarks.
Consumer applications: ChatGPT remains the most-used AI application globally. OpenAI's consumer subscription revenue is a moat that Anthropic does not directly compete with — Claude's consumer presence is growing but remains far smaller.
Enterprise API: Anthropic's revenue is disproportionately driven by API usage and developer adoption. Enterprise teams building AI products tend to benchmark both models and increasingly choose Claude for production workloads requiring reliability and reasoning depth.
Safety and compliance: Anthropic's safety-first positioning — born from its founding mission — gives it an advantage in regulated industries. Government contracts, healthcare, and financial services prefer vendors with credible safety frameworks.
The Business Model Difference
This is the most underappreciated aspect of the rivalry:
OpenAI is a consumer company. ChatGPT subscriptions drive the majority of revenue. The consumer brand is powerful — "ChatGPT" is synonymous with AI the way "Google" was synonymous with search. But consumer businesses are expensive to run: high support costs, churn management, and thin margins on subscription revenue.
Anthropic is an infrastructure company. API revenue and enterprise contracts drive growth. Developers and businesses integrate Claude into their products. This model scales more efficiently — fewer customers at higher contract values, with lower support overhead per dollar of revenue.
The gross margin difference tells the story: Anthropic at 50-77% vs OpenAI at ~33%. Infrastructure companies tend to be more profitable than consumer companies at scale. This is AWS vs Netflix economics.
The People and the Drama
The rivalry is personal. Dario Amodei was VP of Research at OpenAI before leaving in late 2020 with his sister Daniela and seven other senior researchers to found Anthropic. The stated reason: disagreement over OpenAI's direction after the $1 billion Microsoft partnership signaled a shift toward commercialization.
In February 2026, at an AI summit in India, Sam Altman and Dario Amodei visibly refused to hold hands during a group photo. The image went viral — a physical manifestation of a philosophical divide that has only deepened.
The New York Times described the competition as "deeply personal," noting that the Pentagon AI contract controversy involving both companies was just the latest round in a long-running feud. The Wall Street Journal called it "the decadelong feud shaping the future of AI."
What Happens by 2028
Three scenarios, ranked by likelihood:
Most likely — Duopoly (70% probability). Both companies IPO, both reach $100B+ revenue, and the market splits similar to cloud infrastructure. OpenAI dominates consumer. Anthropic dominates developer/enterprise. Neither "wins" outright.
Possible — Anthropic pulls ahead (20% probability). If Anthropic's growth rate holds and its margin advantage compounds, it could become the default AI infrastructure provider. Developer loyalty compounds — once a team builds on Claude, switching costs are high.
Unlikely — OpenAI reasserts dominance (10% probability). A breakthrough model from OpenAI that clearly surpasses Claude across all benchmarks could shift momentum. But the gap in benchmarks is narrow and getting narrower, not wider.
Editorial Opinion
The revenue growth rate is the signal that most people are underweighting. 10x annual growth from a company that is already at billions in revenue is extraordinary. Anthropic is not a startup anymore — it is a hyperscaler that happens to be growing like a startup. The combination of growth rate, margin profile, and developer loyalty is unprecedented.
That said, counting out OpenAI would be foolish. The ChatGPT brand alone is worth billions in customer acquisition cost savings. Consumer mindshare is hard to build and harder to lose. OpenAI's challenge is not demand — it is margin. Generating $30B in revenue while losing money on every token is a different problem than generating $30B with 50% gross margins.
FAQ
Is Anthropic actually bigger than OpenAI now?
By some revenue run rate measures, yes — depending on which report and which month. The gap is close enough that leadership may swap back and forth. What is clear is that the gap has collapsed from 20x to near-parity.
Who has better AI models?
It depends on the task. Claude leads in real-world coding (SWE-bench) and nuanced writing. GPT leads in synthetic benchmarks (HumanEval) and computer use. For most professional use cases, the difference is small enough that other factors — pricing, API reliability, ecosystem — matter more.
Will both companies IPO?
Reports from WSJ and Reuters indicate both are preparing for IPOs. The competitive dynamics may accelerate timing — neither wants to be second to the public markets.
Should I build my product on OpenAI or Anthropic?
For most teams, benchmark both and pick based on your specific use case. For coding-heavy products, Claude's API tends to perform better. For consumer-facing features, GPT's versatility may be advantageous. See our Claude vs ChatGPT comparison for a detailed breakdown.
What about Google (Gemini) and Meta (Llama)?
Both are serious competitors. Google's Gemini 3.1 Pro holds an advantage in context window size (entire monorepos). Meta's open-source Llama models are gaining adoption for self-hosted deployments. The AI market is not a two-player game — but OpenAI and Anthropic are the clear leaders in the proprietary model space.