Anthropic vs OpenAI: How a $47 Billion Run-Rate Flipped the AI Revenue Race

Anthropic's self-reported revenue run-rate hit roughly $47 billion in mid-2026, moving ahead of OpenAI's reported $25–33 billion. Here is what the numbers actually mean, why "run-rate" is not the same as revenue, and how two opposite business models produced the crossover.

For two years the assumption was simple: OpenAI makes ChatGPT, ChatGPT is the most famous product in AI, therefore OpenAI leads on revenue. In mid-2026 that assumption broke. Anthropic — the maker of Claude — reported an annualized revenue run-rate of about $47 billion, while OpenAI's own reported figure sat between $25 billion and $33 billion. The lead changed hands. But the headline hides the more interesting story: the two companies are not really running the same race, and the metric everyone quoted is easy to misread. This post breaks down the figures, the caveat behind "run-rate," and the strategic split that made it happen.

What the numbers actually say

Start with the honest framing. Both companies report annualized run-rate, not audited annual revenue. Run-rate takes a recent period — often a single month — and multiplies it out for a year. It is a forward snapshot of momentum, not a statement of money already earned. For fast-growing companies it flatters the picture, because it projects the best recent month across twelve. Treat every figure below as self-reported and un-audited.

With that caveat, here is the trajectory as reported through mid-2026:

Anthropic's number roughly 4.7x'd in about a year and a half off a 2025 base near $10 billion, and its late-May 2026 disclosure (alongside a large funding round) put the run-rate near $47 billion. OpenAI's reported range of $25–33 billion is still enormous — and still growing — but it was overtaken. The crossover is real even if you discount both figures for run-rate optimism, because Anthropic's growth rate has simply been steeper.

## Why an independent forecast saw it coming

This was not a surprise to everyone. In February 2026, the research group Epoch AI published an analysis of both companies' public revenue trajectories and projected a crossover around August 2026, at roughly $43 billion in annualized revenue. Their fitted trends estimated Anthropic growing near 10x per year since first reporting $1 billion, against OpenAI's roughly 3.4x per year — an enormous gap in slope. Epoch also flagged a caveat that matters: Anthropic's growth may be decelerating, with a possible slowdown to about 7x per year after mid-2025, which would push the crossover later in 2026.

Notably, both companies' own internal forecasts implied slower growth than the raw trend lines. Reporting cited by Epoch suggested OpenAI budgeting around 2.2x growth for 2026 and Anthropic around 4x or less. Even on those more conservative numbers, the models still pointed to a crossover during 2026 or 2027. So the mid-2026 flip landed inside the range independent analysts had already drawn — it was early-to-on-schedule, not an anomaly.

The real story: two opposite business models

The most useful takeaway is not "Anthropic won." It is how the two companies make money, because they earn it in almost mirror-image ways.

• Anthropic is an enterprise/API business. Roughly 85% of its revenue reportedly comes from businesses and developers building on Claude through the API and enterprise contracts. Its growth engine is companies embedding a model into products, coding tools, and back-office workflows — usage that scales with deployment, not with consumer sign-ups.
• OpenAI is a consumer-subscription business. Roughly 85% of its revenue reportedly comes from ChatGPT subscriptions. Its engine is hundreds of millions of individual users, a brand almost synonymous with "AI," and the fastest consumer-product adoption curve in tech history.

That divergence explains the revenue shapes. Enterprise API spend tends to compound quietly: once a company wires Claude into its stack, consumption grows with the company's own activity, contracts renew and expand, and a handful of large customers can move the top line hard. Consumer subscriptions are massive but face a different ceiling — churn, free-tier gravity, and the reality that most people will pay for one AI assistant, not three.

## What it means for the AI race — and for you

Does topping revenue mean Anthropic is "winning" AI? No — and here is the nuance. Revenue leadership is one scoreboard; there are others where the ranking differs. OpenAI still commands the larger consumer footprint and mindshare. Valuations have been a separate contest entirely, with both companies raising at figures in the hundreds of billions. And run-rate leadership can change hands again quickly when growth is this fast and this volatile.

For a reader trying to make sense of it, three practical implications:

1. The enterprise layer is where AI money is quietly concentrating. The company that pulled ahead did it on business spend, not viral consumer growth. If you are watching where durable AI revenue forms, watch API and enterprise deployment, not download charts.
2. Distrust any single "run-rate" headline. It is a momentum metric, self-reported, and un-audited. Useful for direction, unreliable for precision. The trend (Anthropic's steeper slope) is more trustworthy than any one dollar figure.
3. "Winner" depends on the metric. Revenue, users, valuation, and margins can each crown a different leader on the same day. Anyone claiming a single victor is picking a scoreboard.

Frequently Asked Questions

Did Anthropic really pass OpenAI in revenue? On reported annualized run-rate, yes — roughly $47 billion versus OpenAI's reported $25–33 billion in mid-2026. Both figures are self-reported and un-audited, so treat them as directional rather than exact.

What is "annualized run-rate" and why does it matter? It projects a recent short period (often one month) across a full year. It captures momentum but overstates trailing reality for fast-growing firms, which is why it should never be read as audited annual revenue.

Why is Anthropic growing faster? Its revenue is concentrated in enterprise and developer/API usage (about 85%), which compounds as businesses embed Claude into products and workflows. OpenAI leans about 85% on ChatGPT consumer subscriptions, a huge but differently-shaped market.

Did anyone predict this? Yes. Epoch AI projected a crossover around August 2026 near $43 billion in February 2026, based on each company's public revenue trend — Anthropic near 10x/year growth versus OpenAI near 3.4x/year.

Does this make Anthropic the overall leader in AI? Not necessarily. Revenue is one scoreboard. OpenAI still leads on consumer reach and mindshare, and valuations are a separate race. "Leader" depends entirely on which metric you pick.

Key Takeaways

• Anthropic's reported run-rate reached about $47B in mid-2026, moving ahead of OpenAI's reported $25–33B.
• These are self-reported, un-audited run-rate figures — momentum snapshots, not audited annual revenue.
• An independent forecaster, Epoch AI, had projected the crossover around August 2026 near $43B, so the flip was on-schedule.
• The deeper story is model divergence: Anthropic ~85% enterprise/API, OpenAI ~85% consumer subscriptions.
• "Winning" AI depends on the scoreboard — revenue, users, valuation, and margins can crown different leaders.

How this was written Research and a first draft came together with AI's help; verification and the final pass were entirely human.

References

• Anthropic, "Anthropic raises $65B in Series H funding at $965B post-money valuation" (May 28, 2026): https://www.anthropic.com/news/series-h
• Epoch AI, "Anthropic could surpass OpenAI in annualized revenue by mid-2026": https://epoch.ai/data-insights/anthropic-openai-revenue
• CNBC, "Anthropic tops OpenAI as most valuable AI startup, nears $1 trillion valuation in latest round": https://www.cnbc.com/2026/05/28/anthropic-open-ai-startup-value.html
• Simon Willison, "Anthropic's run-rate revenue hits $47 billion": https://simonwillison.net/2026/May/29/anthropic/
• Value Add VC, "OpenAI Revenue 2026: $25B ARR and a -122% Operating Margin": https://valueaddvc.com/blog/openai-revenue-2026-25b-arr-2b-month-and-the-path-to-profitability