On the same day that Anthropic announced its revenue run-rate had crossed $30 billion, OpenAI was sitting at $2 billion in monthly revenue — a $24 billion annualized figure disclosed alongside its record-breaking $122 billion funding round on April 1, 2026. On the face of it, Anthropic has pulled ahead. But the comparison is not that clean. The two companies report revenue differently, and the gap between their headline numbers is partly a product of accounting choices, not just business performance.
The Headline Numbers
Both companies have been on extraordinary trajectories, but their paces have been markedly different.
OpenAI has grown at roughly 3x per year since 2023. Its own CFO Sarah Friar confirmed the numbers in a January 2026 blog post: $2 billion annualized in 2023, $6 billion in 2024, and over $20 billion by the close of 2025. By April 2026, that had grown to $24 billion annualized — $2 billion per month.
Anthropic has grown faster from a smaller base. Its revenue journey went from $100 million in 2023 to $1 billion in 2024, then to $4.5 billion by mid-2025, $9 billion at year-end, $14 billion in February 2026 when it raised its Series G at a $380 billion valuation, and now $30 billion as of early April. That is a 10x annual growth rate, versus OpenAI’s 3x.
Epoch AI, which tracks both companies, had predicted a revenue crossover by mid-2026. Anthropic appears to have arrived a few months early — at least on paper.
How They Calculate Run-Rates (And Why It Matters)
Both companies use annualized run-rate figures rather than reporting audited annual revenue, which is standard practice for high-growth private companies where a single month’s revenue looks very different from the last. The methodology is broadly similar: take four weeks of revenue and multiply by 13. Anthropic adds monthly subscription revenue multiplied by 12 on top of its API run-rate. These are not apples-to-oranges calculations — the approaches are close enough to be comparable.
Where the comparison gets complicated is in how each company handles its cloud distribution partners.
OpenAI reports revenue from its Microsoft Azure partnership on a net basis. Microsoft takes roughly 20% of OpenAI’s revenue from Azure sales. OpenAI deducts that share before recognizing revenue. For direct Azure API sales, OpenAI only counts its cut — not the full amount billed to the end customer.
Anthropic does the opposite. It reports revenue from AWS, Microsoft Azure, and Google Cloud on a gross basis — booking the full amount billed through those channels, including the cloud providers’ share, as its top-line revenue. The cloud partners’ portions are then listed as sales and marketing costs on the expense side.
Both approaches are permissible under US GAAP (the relevant accounting standard is ASC 606, which hinges on whether a company is acting as the “principal” or “agent” in a sale). Anthropic positions itself as the primary provider; OpenAI treats Microsoft as the primary provider for Azure. Both are defensible accounting positions. But they produce very different revenue headlines for economically similar transactions.
The practical implication: if Anthropic reported revenue the same way OpenAI does, its top-line number would be materially lower. Analysts at The Information flagged that on a comparable basis, Anthropic’s reported revenue would be “meaningfully lower” — though neither company discloses enough to calculate the precise difference.
This distinction will matter enormously when both companies eventually file S-1 prospectuses for their anticipated IPOs. Public market investors will apply valuation multiples to revenue, and a company reporting gross revenue from cloud resales will show a higher top line and lower gross margin than an economically equivalent company reporting net. The accounting methodology becomes the valuation story.
Historical Run-Rate Comparison
| Period | OpenAI | Anthropic |
|---|---|---|
| 2023 (full year) | ~$2B | ~$100M |
| 2024 (full year) | ~$6B | ~$1B |
| Mid-2025 | ~$12–13B | ~$4.5B |
| End of 2025 | ~$20B+ | ~$9B |
| February 2026 | ~$25B | ~$14B |
| April 2026 | ~$24B | ~$30B |
Note: All figures are annualized run-rates, not audited annual revenue. Anthropic reports gross; OpenAI reports net for cloud partner sales.
The table shows Anthropic closing the gap at an aggressive pace throughout 2025 and into 2026 — and the crossover arriving in early April 2026. OpenAI’s run-rate actually appeared to dip slightly between February and April (from ~$25B to ~$24B on a strict annualized basis), possibly reflecting the quarterly rhythm of enterprise renewals or a slower month.
The Accounting Question Won’t Stay Abstract
The gross-versus-net issue is not just an accounting footnote. Anthropic’s enterprise distribution is heavily weighted toward cloud marketplaces — AWS Bedrock and Google Cloud Vertex AI account for a significant share of how enterprise customers access Claude. The cloud providers take a percentage of those sales. Under Anthropic’s current reporting methodology, that full gross amount flows through its revenue line.
OpenAI, by contrast, built much of its business on direct subscriptions through ChatGPT — a consumer product where Microsoft takes no cut — and API sales that flow through Azure at the net rate. Its revenue quality, in gross margin terms, is arguably cleaner.
As both companies approach public markets, the accounting methodology embedded in each S-1 will define how investors benchmark them. A revenue multiple applied to Anthropic’s gross run-rate implies a very different valuation than the same multiple applied to a net equivalent.
What Is Clear, And What Isn’t
What is clear: both companies are growing at speeds that have no real precedent in enterprise technology. OpenAI went from $1 billion in annual revenue to $24 billion annualized in roughly three years. Anthropic did the same journey in less time, from a smaller base, and at a faster rate. Epoch AI has estimated that Anthropic’s 10x annual growth rate dwarfs OpenAI’s 3.4x, and the data through April 2026 bears that out.
What is not clear: whether Anthropic’s $30 billion is genuinely larger than OpenAI’s $24 billion in economic terms, or whether the accounting difference accounts for a meaningful portion of the gap. Neither company discloses the hyperscaler revenue splits in sufficient detail to calculate a precise like-for-like comparison.
The honest answer to “has Anthropic passed OpenAI?” is: probably yes in gross terms, probably not yet in net terms, and the truth lies somewhere in between. For now, both numbers will keep moving — and likely faster than anyone is forecasting.