The latest OpenAI vs Anthropic statistics show the AI duopoly inverted in April 2026, when Anthropic crossed approximately $30 billion in annualized revenue run-rate just two months after OpenAI‘s end-of-February $25 billion run-rate, the first time a frontier-AI challenger eclipsed the category incumbent on reported revenue. By late May, Anthropic’s Series H raised $65 billion at approximately a $965 billion post-money valuation, briefly putting it ahead of OpenAI’s $852 billion post-money in private markets too.
One caveat readers should carry through every figure: the “ARR” tag on AI run-rates compresses contracted ARR (CARR), recognized ARR, and annualized run-rate extrapolations into one headline. Per TechCrunch’s May 2026 reporting, the CARR-vs-ARR measurement gap can reach 70% at AI startups, and Sacra’s May estimate sits above either company’s formal disclosure.
Key Takeaways
- Anthropic’s annualized run-rate climbed from roughly $87 million in January 2024 to about $30 billion by April 2026, an expansion the company attributes to enterprise API demand.
- OpenAI’s annualized revenue moved from $21.4 billion at year-end to $25 billion by the end of February 2026, a 17% gain over two months per The Information.
- Anthropic’s enterprise customer cohort spending more than $1 million annually doubled in recent months to exceed 1,000 clients by April 2026, driving the majority of run-rate growth.
- ChatGPT serves more than 900 million people weekly, while Anthropic counts 300,000+ business customers generating roughly 80% of revenue through pay-per-token API access.
- Ramp’s May 2026 AI Index recorded Anthropic at 34.4% versus OpenAI at 32.3% of enterprise corporate-card AI spend, the first such reversal.
- Microsoft’s revised partnership with OpenAI values its stake at roughly $135 billion, or 27% of the for-profit on an as-converted diluted basis, down from 32.5% before the recapitalization.
- Anthropic’s Claude Opus 4.8 scored 69.2% on SWE-bench Pro versus OpenAI’s GPT-5.5 scoring 84.9% on GDPval, reflecting divergent benchmark strengths.
Editor’s Choice
- Anthropic confidentially filed an IPO S-1 on June 1, 2026, at a $965 billion valuation.
- OpenAI closed a $122 billion funding round on March 31, 2026, at an $852 billion post-money valuation.
- Sacra estimates Anthropic reached $47 billion in annualized revenue by May 2026, up from $9 billion at year-end 2025.
- OpenAI’s projected cash burn is approximately $27 billion in 2026 and roughly $63 billion in 2027.
- Anthropic’s Claude Code product hit $2.5 billion in annualized revenue by February 2026, up from $1 billion in November 2025.
- OpenAI’s inference costs reached $8.4 billion in 2025 and are projected to reach $14.1 billion in 2026.
Recent Developments
- June 1, 2026: Anthropic confidentially submitted a draft S-1 to the SEC at a $965 billion valuation, with the IPO tentatively planned for October 23, 2026, on NASDAQ.
- May 28, 2026: Anthropic closed a $65 billion Series H led by Altimeter Capital, Dragoneer, Greenoaks, and Sequoia.
- May 28, 2026: Anthropic launched Claude Opus 4.8 with 69.2% on SWE-bench Pro and 88.6% on SWE-bench Verified, priced unchanged at $5 per million input tokens and $25 per million output tokens.
- May 2026: Ramp’s AI Index recorded Anthropic at 34.4% of enterprise corporate-card AI spend versus OpenAI’s 32.3%, the first such reversal.
- April 27, 2026: OpenAI and Microsoft announced a revamped partnership letting OpenAI sell across any cloud provider, including Amazon and Google.
- April 23, 2026: OpenAI released GPT-5.5, the first fully retrained base model since GPT-4.5, scoring 84.9% on GDPval.
- April 7, 2026: Anthropic announced about a $30 billion annualized run-rate, roughly doubling from $14 billion in February.
OpenAI vs Anthropic Revenue Run-Rates
OpenAI and Anthropic disclose annualized run-rates rather than audited GAAP revenue, so direct comparison requires noting each disclosure date.
OpenAI’s reported figure of $25 billion at end-February 2026 is the most-recent OpenAI-attributable run-rate in published reporting; The Information sourced it, and Reuters could not independently verify. Anthropic’s roughly $30 billion April 2026 reading sits in a different reporting window, and Sacra’s later $47 billion May 2026 estimate is the analyst firm’s forward extrapolation rather than a company disclosure.
The trajectory matters more than any single snapshot. Anthropic’s run-rate climbed from roughly $87 million in January 2024 to approximately $30 billion by April 2026, a pace VentureBeat described as the company crossing $30 billion in under three years from a standing start.
By the numbers: Anthropic’s revenue trajectory moved from roughly $87 million (January 2024), to $1 billion (December 2024), to about $9 billion (end of 2025), to $30 billion (April 2026). The same period saw OpenAI grow from roughly $20 billion at year-end 2025 to $25 billion by February 2026, a 25% gain against Anthropic’s 233% over the same window.
Valuation Trajectory
OpenAI’s March 31, 2026 round raised $122 billion at an $852 billion post-money valuation from a syndicate co-led by SoftBank with Andreessen Horowitz, D.E. Shaw Ventures, MGX, TPG, and T. Rowe Price-advised accounts. Strategic participation came from Amazon, NVIDIA, and Microsoft. Anthropic’s Series H closed on May 28, 2026, at $65 billion raised and approximately $965 billion post-money, led by Altimeter Capital, Dragoneer, Greenoaks, and Sequoia Capital.
The Series H round eclipsed OpenAI’s private valuation for the first time. Pre-IPO Anthropic shares traded at roughly $572 per share on secondary markets as of late May 2026, according to Fortune. The valuation gap of $113 billion is roughly half of Salesforce, and it reverses three years of consensus that OpenAI commanded the category premium.
Enterprise Customer Base and Spend Share
The enterprise spend inversion happened quickly. According to Ramp’s reading, Anthropic captured over 73% of all spending among companies purchasing AI tools for the first time in March 2026, reversing a 60/40 OpenAI advantage as recently as early December 2025 and a 50/50 split in January 2026. Ramp’s data covers its own customer base, so the figures index enterprise behavior on Ramp’s platform rather than the full corporate market.
| Metric | OpenAI | Anthropic |
|---|---|---|
| Enterprise customers paying over $1 million per year | not disclosed | 1,000+ (April 2026) |
| Total business customers | 9 million paying business users (Feb 2026) | 300,000+ (Oct 2025) |
| Share of revenue from enterprise | 40% (parity with consumer by end-2026) | ~80% |
| Pricing model | Per-seat (ChatGPT Enterprise) + API | Pay-per-token API + Claude Code |
| Bedrock / multi-cloud | Recently expanded April 2026 | 100,000+ on Amazon Bedrock (April 2026) |
Source: Sacra, VentureBeat, OpenAI announcements
Anthropic’s 300,000+ business customers generate approximately 80% of revenue through pay-per-token API access, with more than 100,000 of those running Claude on Amazon Bedrock as of April 2026. The high-end cohort scaled fastest: customers spending more than $100,000 annually grew seven-fold year over year, and customers spending more than $1 million annually more than doubled in recent months to exceed 1,000 by April 2026.
OpenAI’s enterprise mix represents more than 40% of revenue and is on track to reach parity with consumer revenue by end of 2026. Sacra counted more than 9 million paying business users as of February 2026 across ChatGPT Team, Enterprise, and Edu, alongside roughly 15 million ChatGPT Plus consumer subscribers as of mid-2025.
Why it matters: Anthropic’s enterprise revenue concentration (~80% of total) means the company depends on API monetization and developer tooling, while OpenAI’s roughly 60/40 consumer-enterprise split makes ChatGPT’s more than 900 million weekly user base the dominant exposure. Different revenue mixes imply different competitive risks.
Funding Rounds and Cap Table
| Round | OpenAI | Anthropic |
|---|---|---|
| Most recent raise | $122 billion (March 31, 2026) | $65 billion (May 28, 2026) |
| Post-money valuation | $852 billion | $965 billion |
| Round lead | SoftBank co-led | Altimeter Capital |
| Prior round | n/a (PBC restructure Oct 28, 2025) | $30 billion Series G at $380 billion (Feb 12, 2026) |
| Strategic investors | Amazon, NVIDIA, Microsoft | not individually disclosed |
| Credit facility | $4.7 billion revolving (undrawn at close) | not disclosed |
Source: OpenAI announcement, Anthropic newsroom, Fortune
The Series G to Series H jump tells the speed story for Anthropic. The February 12, 2026 Series G raised $30 billion at a $380 billion post-money valuation; by May 28, the Series H added $65 billion at approximately $965 billion post-money, a valuation increase of roughly 2.5x over fifteen weeks. OpenAI’s simultaneous approximately $4.7 billion revolving credit facility was undrawn at close, leaving substantial dry powder against the company’s roughly $600 billion total compute spend target through 2030.
Worth noting: Anthropic’s Series H eclipsed OpenAI’s private valuation for the first time, but the round was tied directly to the company’s confidential S-1 filing three days later. Public-market dynamics could compress the spread quickly.
Model Benchmarks: Claude Opus 4.8 vs GPT-5.5
| Benchmark | Claude Opus 4.8 | Claude Opus 4.7 | GPT-5.5 |
|---|---|---|---|
| SWE-bench Pro (coding) | 69.2% | 64.3% | not directly comparable |
| SWE-bench Verified | 88.6% | 87.6% | not disclosed |
| MCP-Atlas | 82.2% | 77.3% | n/a |
| BrowseComp (single-agent) | 84.3% | 79.3% | n/a |
| GDPval (knowledge work) | not disclosed | not disclosed | 84.9% |
| OSWorld-Verified | not disclosed | not disclosed | 78.7% |
| Tau2-bench Telecom | not disclosed | not disclosed | 98.0% |
| Terminal-Bench 2.0 | not disclosed | not disclosed | 82.7% |
| ARC-AGI-2 | not disclosed | not disclosed | 85.0% |
| MRCR v2 (long-context) | not disclosed | not disclosed | 74.0% |
Source: Anthropic Claude Opus 4.8 announcement, OpenAI GPT-5.5 announcement
The two model lines now compete on different benchmarks more than on a shared scoreboard. Anthropic launched Opus 4.8 on May 28, 2026, scoring 69.2% on SWE-bench Pro, a 4.9-point gain over Opus 4.7. An Originality.ai AI study also found that Claude Opus 4.7 remains highly detectable as AI-generated, offering a practical perspective on the model beyond benchmark performance. The model also scored 88.6% on SWE-bench Verified, 82.2% on MCP-Atlas, and 84.3% on BrowseComp single-agent. Anthropic’s release note added that Opus 4.8 is approximately four times less likely than its predecessor to allow flaws in its own generated code to pass unremarked, and that pricing held at $5 per million input tokens and $25 per million output tokens.
OpenAI’s GPT-5.5 launched April 23, 2026, as the first fully retrained base model since GPT-4.5. The model scored 84.9% on GDPval, a benchmark testing knowledge work across 44 occupations, and 78.7% on OSWorld-Verified, which measures whether a model can operate real computer environments. On Tau2-bench Telecom, it reached 98.0% without prompt tuning; on Terminal-Bench 2.0 it scored 82.7% versus 69.4% for the prior model; on ARC-AGI-2 85.0% versus 75.8%, and on long-context retrieval (MRCR v2) 74.0% versus 32.2%.
Anthropic leads on coding evaluations and OpenAI leads on agentic-knowledge-work suites; the next release cycle will likely scramble that. GPT-5.5 Instant rolled out to ChatGPT free-tier users on May 5, 2026, replacing GPT-5.3 Instant as the default model.
Claude Code and ChatGPT API Adoption
- Anthropic’s Claude Code reached $2.5 billion in annualized revenue by February 2026, up from $1 billion in November 2025.
- Software development represents approximately 37% of business-context Claude conversations per Anthropic’s March 2026 Economic Index.
- ChatGPT serves more than 900 million people weekly; OpenAI counts more than 9 million paying business users as of February 2026.
Anthropic’s developer monetization runs through two surfaces: the raw API and the standalone Claude Code product. Claude Code became generally available in May 2025, hit $1 billion in annualized revenue by November 2025, and reached $2.5 billion in annualized revenue by February 2026. Enterprise deployments accounted for more than half of Claude Code’s revenue base. Anthropic’s March 2026 Economic Index reported that software development represented approximately 37% of business-context Claude conversations, underlining the developer skew.
OpenAI’s distribution looks different. ChatGPT serves more than 900 million people weekly, with roughly 15 million ChatGPT Plus subscribers at $20 per month as of mid-2025 and more than 9 million paying business users as of February 2026 across ChatGPT Team, Enterprise, and Edu.
| Metric | OpenAI | Anthropic |
|---|---|---|
| Weekly active users (consumer + business) | 900 million+ | not disclosed |
| Paid consumer subscribers | ~15 million Plus (mid-2025) | not disclosed |
| Paid business users | 9 million+ (Feb 2026) | not disclosed individually |
| Business customers | not separately disclosed | 300,000+ (Oct 2025) |
| Dedicated coding product | none branded standalone | Claude Code, $2.5 billion ARR (Feb 2026) |
Source: Sacra, OpenAI announcement, Anthropic Economic Index
The takeaway: OpenAI’s lead in weekly active reach is unambiguous. Anthropic’s lead in revenue per business customer reflects a deliberately developer-skewed product surface. Both companies grew their named-customer counts in 2026, but the unit economics differ enough that direct user-count comparisons miss the financial story.
Microsoft Partnership and OpenAI’s PBC Restructure
- Microsoft’s stake in OpenAI’s for-profit arm sits at roughly 27% on an as-converted diluted basis, valued at about $135 billion.
- The OpenAI Foundation (the nonprofit) holds an equity stake worth approximately $130 billion in the for-profit arm.
- Microsoft’s prior stake was 32.5% before the October 28, 2025 recapitalization.
- OpenAI’s April 27, 2026 revamped agreement allows the company to sell across any cloud provider, including Amazon and Google.
- Microsoft no longer pays a revenue share to OpenAI; OpenAI’s payments to Microsoft continue until 2030, subject to an overall cap.
OpenAI’s corporate structure changed materially in late 2025 and early 2026. On October 28, 2025, OpenAI completed its recapitalization as a Public Benefit Corporation called OpenAI Group PBC, with the nonprofit (renamed the OpenAI Foundation) holding an equity stake worth about $130 billion in the for-profit arm. Microsoft’s investment was valued at $135 billion, or roughly 27% of the company on an as-converted diluted basis, down from a previous 32.5% stake before the recapitalization.
The partnership restructure followed six months later. On April 27, 2026, OpenAI and Microsoft announced a revamped agreement permitting OpenAI to cap revenue-share payments and serve customers across any cloud provider, including Amazon and Google. Under the new terms, Microsoft no longer pays a revenue share to OpenAI; payments in the opposite direction continue until 2030, with OpenAI’s share-of-revenue obligation now independent of technological progress and subject to an overall cap.
Worth noting: The cap and multi-cloud permission convert OpenAI’s distribution from Microsoft-Azure-exclusive to broadly accessible, part of why Microsoft accepted dilution from 32.5% to 27% in exchange.
Anthropic’s capital structure is conventional in comparison. The company is privately held with a confidential S-1 filed on June 1, 2026, ahead of a planned October 23, 2026, NASDAQ listing. Amazon and Google are Anthropic’s largest strategic investors, but neither approaches a 27% stake comparable to Microsoft’s OpenAI position.
How “ARR” Is Computed: A Measurement Caveat
- TechCrunch reported in May 2026 that the contracted-vs-recognized ARR gap can reach 70% at AI startups.
- One startup publicly claimed $50 million in ARR while actual figures showed $42 million.
- Another high-profile enterprise startup reported over $100 million in ARR with only a fraction coming from currently paying customers.
- Annualized run-rate extrapolations can be computed from any window, including a quarter, month, week, or even a day.
The unique angle to surface in any OpenAI versus Anthropic comparison is that the headline “ARR” figures both companies report use different methodologies. TechCrunch reported in May 2026 that AI startups commonly substitute contracted ARR (CARR) for traditional ARR, with one investor saying that “for sure they are reporting CARR as ARR.” A venture capitalist observed companies where contracted ARR exceeded recognized ARR by 70% higher than ARR, with much of that contracted revenue likely never materializing. The same TechCrunch piece quoted Spellbook co-founder Scott Stevenson calling the practice a huge scam in public commentary.
Three measurement conventions coexist in the current reporting:
- Recognized ARR, revenue from already-onboarded customers, the most conservative reading. This is closest to GAAP.
- Contracted ARR (CARR) adds revenue from signed customers not yet onboarded, which inflates the number when implementation timelines slip or contracts cancel.
- Annualized run-rate revenue extrapolates current revenue across the next twelve months based on a recent period’s haul, which the article notes can be calculated from a quarter, month, week, or even a day.
Because many AI companies charge based on usage or outcomes, the annualized run-rate is no longer locked into predictable contracts and can swing meaningfully on a single large prompt-volume month. A given headline of “$30 billion ARR” may compress all three definitions; readers comparing OpenAI’s $25 billion run-rate to Anthropic’s $30 billion are comparing reporting conventions as much as steady-state revenue.
The 2026 revenue figures from both companies are reliable as growth indicators but should not be read as GAAP-equivalent. The genuine answer to “who is bigger” requires waiting for Anthropic’s public S-1 disclosure and OpenAI’s own structured financial reporting.
Is Anthropic worth more than OpenAI?
By post-money valuation as of late May 2026, yes. Anthropic’s Series H closed at approximately $965 billion post-money on May 28, 2026, while OpenAI’s March 31, 2026 round closed at $852 billion post-money, a $113 billion gap in Anthropic’s favor. Both readings are private secondary markings and may not survive an IPO price discovery, but on the available data, Anthropic carries the higher valuation today.
Why is OpenAI valued more than Anthropic?
It is no longer. For most of 2023 through 2025, OpenAI carried the higher valuation because of ChatGPT’s consumer scale and Microsoft’s commercial commitment. That premium reversed in May 2026 when Anthropic’s Series H eclipsed OpenAI’s private valuation for the first time. The reasons sit in enterprise momentum: a 233% revenue gain over the same window that produced OpenAI’s 25% gain, plus a higher share of enterprise spend per Ramp, and a developer-led product surface that compounds at usage-based pricing.
Who is winning, Anthropic or OpenAI?
Anthropic and OpenAI win on different dimensions in the current snapshot. Anthropic leads on reported revenue run-rate (about $30 billion in April vs $25 billion in February), post-money valuation (approximately $965 billion vs $852 billion), and enterprise corporate-card spend share (34.4% vs 32.3% in May). OpenAI leads on consumer reach (more than 900 million weekly users) and on certain agentic-benchmark suites including Tau2-bench Telecom (98.0%) and Terminal-Bench 2.0 (82.7%). Calling a winner depends on the metric. Anthropic carries the momentum; OpenAI carries the breadth.
Conclusion
The 2026 numbers reshape an AI duopoly that looked uncontestable a year ago. Anthropic’s about $30 billion April annualized run-rate, approximately $965 billion Series H post-money, and 34.4% versus 32.3% enterprise spend share mark the first time a frontier-AI challenger has surpassed OpenAI on three financial dimensions simultaneously, while Claude AI monetizes through API depth.
OpenAI continues to monetize through consumer reach. Anthropic’s confidential S-1 filing and planned October NASDAQ listing make the next quarter the most consequential moment in private-market AI since OpenAI’s $122 billion close, and if the SEC-filed financials hold up the run-rate numbers under audit, the 2026 inversion will look durable.
