Is Anthropic now ahead of OpenAI?

What metric should we use to compare OpenAI and Anthropic?

The cleanest way to compare OpenAI and Anthropic is revenue run-rate, not users.

User metrics look tempting because OpenAI has unusually strong public data: ChatGPT weekly active users, paying subscribers, and app-style estimates.

We can trace ChatGPT from 100M weekly active users in late 2023 to 900M by early 2026, giving OpenAI a clean adoption curve.

But Anthropic does not report Claude-wide usage the same way. It shares narrower signals, especially for Claude Code, such as run-rate revenue or weekly active user growth.

Comparing ChatGPT WAU to Claude Code WAU would mix a full consumer product with a coding product. Comparing ChatGPT MAU to third-party Claude traffic estimates would also mix reported data with outside estimates.

Paying users are messy too. A ChatGPT Plus subscriber, Claude Code user, API customer, and seven-figure enterprise contract all count as “paid usage,” but reflect very different monetization models.

That makes ARR or revenue run-rate the best apples-to-apples metric. It is not audited annual revenue, but it is the cleanest available signal for current business scaling.

If you want more recent data on this point, please see our latest generative AI market report.

What has OpenAI and Anthropic communicated about ARR and revenue run-rate so far?

Frankly, OpenAI gives us a clean annual ARR series, while Anthropic gives us more frequent run-rate checkpoints.

OpenAI’s public numbers show a very large business compounding year by year: $2B ARR in 2023, $6B in 2024, and $20B+ in 2025.

Anthropic’s public numbers show a more compressed acceleration curve: roughly $1B run-rate at the beginning of 2025, $5B+ by August 2025, $14B by February 2026, and $30B+ by April 2026.

That means OpenAI still has the cleaner multi-year scale story, but Anthropic has the more dramatic recent velocity story.

Date	Company	What was communicated	Comment
April 2026	Anthropic	$30B+ revenue run-rate	This was the sharpest jump in the sequence. Anthropic said it had moved from roughly $9B at the end of 2025 to $30B+ in early April 2026, more than tripling in about one quarter. The previous comparable OpenAI disclosure was $20B+ ARR for 2025, so this was the first public company-reported point where Anthropic’s run-rate appeared to move above OpenAI’s last disclosed ARR level.
February 2026	Anthropic	$14B revenue run-rate	Anthropic framed this as less than three years after earning its first dollar of revenue. The important detail is not just the number, but the customer mix: more than 500 customers were spending over $1M annualized, and Claude Code alone was above $2.5B run-rate. Compared with OpenAI’s $20B+ ARR for 2025, Anthropic was still smaller on the latest clean OpenAI number, but catching up fast.
January 2026	OpenAI	$20B+ ARR for 2025	OpenAI’s CFO gave the cleanest OpenAI revenue series: $2B in 2023 → $6B in 2024 → $20B+ in 2025. That is about 10x in two years. The interesting interpretation is that OpenAI explicitly linked revenue to compute availability, meaning demand was not the only growth variable; supply constraints also shaped how fast revenue could be realized.
End of 2025	Anthropic	~$9B revenue run-rate	Anthropic later used this as the baseline for its April 2026 jump to $30B+. This is the bridge number that matters: by the end of 2025, Anthropic was already above OpenAI’s 2024 ARR of $6B, but still far below OpenAI’s 2025 ARR of $20B+. The gap then narrowed dramatically in the following quarter.
August 2025	Anthropic	$5B+ revenue run-rate	Anthropic said it had grown from roughly $1B at the beginning of 2025 to $5B+ by August. This is where the curve starts to look different from a normal SaaS ramp. Claude Code was already at $500M+ run-rate, showing that coding was becoming a material growth engine, not just a feature narrative.
2024	OpenAI	$6B ARR	OpenAI tripled from $2B ARR in 2023 to $6B ARR in 2024. Against Anthropic’s later $1B early-2025 baseline, OpenAI was still materially ahead at this stage. This was the phase where ChatGPT’s consumer scale, API usage, and business subscriptions began converting into a real revenue engine.
Beginning of 2025	Anthropic	~$1B revenue run-rate	Anthropic’s early-2025 run-rate shows how steep the later curve became. From this base, it reached $5B+ by August 2025, ~$9B by year-end, and $30B+ by April 2026. In other words, Anthropic’s revenue story is less about long historical scale and more about sudden enterprise/developer compression.
2023	OpenAI	$2B ARR	This is the first point in OpenAI’s clean ARR series. It came after ChatGPT had already become a mass-market product, so the key read is that OpenAI monetized distribution very early. By comparison, Anthropic’s first disclosed billion-dollar run-rate came later, but then accelerated much faster through enterprise and developer adoption.

Has OpenAI always been ahead of Anthropic on revenue scale?

OpenAI was publicly ahead of Anthropic on revenue scale from the first comparable signals through February 2026, but the public record flips in April 2026, when Anthropic disclosed a $30B+ revenue run-rate, roughly 1.5x OpenAI’s latest disclosed $20B+ ARR.

That is the precise answer. OpenAI led the first phase by a wide margin. It was founded in 2015, launched ChatGPT in November 2022, and reached $2B ARR in 2023. Anthropic was founded in 2021, launched Claude publicly in 2023, and only later disclosed a roughly $1B revenue run-rate around the beginning of 2025.

We do not have a clean Anthropic 2023 revenue number, but based on the public sequence, OpenAI was almost certainly several times larger at that point.

By 2024, the gap was still very large. OpenAI reported $6B ARR for 2024, while Anthropic’s next disclosed baseline was roughly $1B run-rate at the beginning of 2025. That suggests OpenAI was about 6x larger around that transition point. This is the period where OpenAI’s advantage was not subtle: it had the bigger product, the bigger user base, and the bigger revenue engine.

The gap compressed hard in 2025. OpenAI reached $20B+ ARR for 2025, while Anthropic moved from roughly $1B at the beginning of the year to $5B+ by August and around $9B by year-end. So by the end of 2025, Anthropic had already passed OpenAI’s 2024 level, but OpenAI was still ahead on the latest comparable public numbers: $20B+ versus roughly $9B, or more than 2x larger.

That was still true in February 2026. Anthropic disclosed a $14B revenue run-rate, which was a huge jump, but still below OpenAI’s latest disclosed $20B+ ARR. On the public numbers available then, OpenAI was still ahead by at least 1.4x, assuming OpenAI had not reported a newer figure.

The crossover appears in April 2026. Anthropic then disclosed a $30B+ revenue run-rate, up from roughly $9B at the end of 2025. That was the first company-reported number above OpenAI’s latest disclosed $20B+ ARR.

To make things super clear: OpenAI was ahead at $20B+ in 2025; Anthropic was still behind at $14B in February 2026; then Anthropic moved ahead publicly at $30B+ in April 2026.

If you want more recent data on this point, please see our latest generative AI market report.

Period	OpenAI revenue scale	Anthropic revenue scale	Gap / read
2023	$2B ARR	No clean disclosed number	OpenAI was clearly ahead. Anthropic had launched Claude publicly, but we do not have a comparable revenue disclosure yet.
2024 → early 2025	$6B ARR	~$1B run-rate	OpenAI was about 6x larger around this transition point. This was still a wide-gap phase.
August 2025	Latest disclosed: $20B+ ARR for 2025	$5B+ run-rate	OpenAI was still at least ~4x larger using the latest public OpenAI figure. Anthropic was accelerating, but still behind.
End of 2025	$20B+ ARR	~$9B run-rate	OpenAI was still more than 2x larger, but Anthropic had already passed OpenAI’s 2024 level.
February 2026	Latest disclosed: $20B+ ARR	$14B run-rate	OpenAI was still ahead by at least ~1.4x, assuming no newer OpenAI number.
April 2026	Latest disclosed: $20B+ ARR	$30B+ run-rate	Public record flips: Anthropic’s disclosed run-rate was roughly 1.5x above OpenAI’s last disclosed ARR.

How fast did Anthropic close the revenue gap with OpenAI?

Anthropic caught up to OpenAI’s public revenue surprisingly quickly: from roughly 6x smaller around the 2024 to early-2025 transition, to more than 2x smaller by the end of 2025, to apparently ahead by April 2026.

That compression is the interesting part. Around the 2024 to early-2025 transition, OpenAI had disclosed $6B ARR, while Anthropic was around $1B run-rate. By the end of 2025, OpenAI’s latest disclosed number was $20B+ ARR, while Anthropic was around $9B run-rate.

Then, in February 2026, Anthropic reached $14B run-rate, still below OpenAI’s latest disclosed $20B+ figure. Two months later, Anthropic disclosed $30B+ run-rate, which was roughly 1.5x above OpenAI’s last public number.

Anthropic did not need ChatGPT-scale consumer distribution to close the revenue gap. It found denser monetization: enterprise demand, API workloads, cloud distribution, and especially coding workflows.

Is Anthropic bigger than OpenAI now?

Anthropic shows an impressive $30B+ run-rate but it does not prove it is definitively bigger than OpenAI. It proves that Anthropic’s latest disclosed run-rate was above OpenAI’s latest disclosed ARR.

That distinction matters. A run-rate is a current annualized pace, not audited full-year revenue. So Anthropic’s April 2026 figure does not prove it generated more full-year revenue than OpenAI. It only tells us that, at that point in the public record, Anthropic had disclosed a higher current revenue pace than OpenAI’s last disclosed figure.

OpenAI may already have grown beyond $20B+ by then. But unless OpenAI reports a newer number, we cannot put that into the chronology.

If you want more recent data on this point, please see our latest generative AI market report.

How did that happen? Why is Anthropic beating OpenAI now?

Anthropic did not close the revenue gap because it became “the new ChatGPT.” It closed the gap because it monetized professional usage much harder than its consumer footprint suggests.

OpenAI was not falling apart. It went from $2B ARR in 2023 to $6B in 2024 and $20B+ in 2025, so roughly 10x in two years. That is already insane growth. The problem is that Anthropic’s curve was even more violent: roughly $1B run-rate at the beginning of 2025, $5B+ by August 2025, ~$9B by year-end, $14B in February 2026, and $30B+ in April 2026. That is about 30x in roughly fifteen months.

The first reason is simple: Anthropic’s customers got big fast. By August 2025, it had 300,000+ business customers, and accounts worth $100K+ in run-rate revenue had grown nearly 7x YoY. By February 2026, it had 500+ customers spending $1M+ annualized. By April 2026, it had 1,000+. That is the cleanest signal in the whole story: Anthropic doubled its seven-figure customer count in less than two months.

The second reason is Claude Code. This is where the story stops being vague. Claude Code launched publicly in May 2025. By August 2025, it was already above $500M run-rate, with usage up more than 10x in three months. By February 2026, it was above $2.5B run-rate. Anthropic also said Claude Code weekly active users had doubled since January 1, business subscriptions had quadrupled, and enterprise usage represented more than half of Claude Code revenue.

And actually, coding is not a casual chatbot use case. It is one of the easiest AI budgets to justify. If Claude saves engineering hours, speeds up migrations, fixes bugs, or helps ship code faster, companies can pay serious money for it. A consumer might pay $20 a month. An engineering org can spend six or seven figures a year if the ROI is obvious.

The third reason is that Claude was sold where enterprises already buy AI. Anthropic did not need every user to go to Claude.ai. Claude was available through AWS Bedrock, Google Cloud Vertex AI, and Microsoft Azure Foundry. That means it could flow through existing cloud contracts, security reviews, procurement processes, and infrastructure budgets. Boring distribution, massive impact.

The fourth reason is workload density. Claude was not just being used for general chat. The revenue came from coding, API workloads, data analysis, automation, financial analysis, legal work, cybersecurity, and scientific workflows. These are high-cost labor markets. If the model works, the buyer does not think in “$20 subscription” terms. They think in saved headcount, faster projects, lower consulting spend, and more output per team.

The fifth reason is compute. Anthropic tied its $30B+ April 2026 run-rate to customer demand and expanded compute with Google TPUs, AWS Trainium, and Nvidia GPUs. In AI, demand is not enough. You only book the revenue if you can actually serve the workloads. More capacity meant more demand could become billable usage.

So the real answer is not “Anthropic won enterprise.” That is too lazy. The sharper answer is: Anthropic found a higher-revenue slice of the AI market. OpenAI had the bigger consumer audience. Anthropic had a denser monetization loop: professional users, engineering teams, cloud buyers, API workloads, and seven-figure enterprise accounts.

This was not a consumer adoption story but a budget reallocation story. Companies moved money from developer tools, consulting hours, internal automation projects, cloud AI experiments, and software seats into Claude usage. That is how Anthropic closed the gap without needing anything close to ChatGPT’s consumer scale.

Driver	What happened / why it mattered
Enterprise accounts	$1M+ customers grew from 500+ to 1,000+ in under two months, directly lifting revenue density.
Claude Code	Reached $500M+ run-rate in Aug. 2025, then $2.5B+ by Feb. 2026, becoming a major growth engine.
Cloud distribution	Availability through AWS, Google Cloud, and Azure made enterprise procurement easier.
High-ROI workflows	Coding and API workloads justify large budgets because they save expensive engineering time.
Compute capacity	More compute let Anthropic convert demand into revenue instead of being supply-constrained.

If you want more recent data on this point, please see our latest generative AI market report.

Did people abandon ChatGPT for Claude?

No, people did not abandon ChatGPT for Claude. Anthropic’s catch-up was not a consumer exodus; it was a professional spending story.

The numbers make that obvious. In February 2026, ChatGPT still had 900M weekly active users and 50M paying subscribers. Claude was growing, but third-party app estimates put it around 56M monthly active app users in Q2 2026, versus roughly 1B for ChatGPT. That means Claude was still about 18x smaller on consumer app usage.

So no, the story is not “everyone left ChatGPT.” If that had happened, we would see Claude catching ChatGPT in app users, paid subscribers, or mainstream consumer usage. We do not. ChatGPT kept the consumer attention lead by a huge margin.

The thing is … Claude monetized professional usage much harder. Anthropic went from 300,000+ business customers in August 2025 to 500+ customers spending $1M+ annualized in February 2026, then 1,000+ $1M+ customers by April 2026. That is not a consumer migration. That is enterprise expansion.

Claude Code proves the point. It launched publicly in May 2025, reached $500M+ run-rate by August 2025, then passed $2.5B+ run-rate by February 2026. That is coding, API, and workplace usage turning into real budget fast.

The important distinction is simple: ChatGPT had more people; Claude had more revenue density in high-value workflows. A casual consumer user may pay nothing. A ChatGPT Plus user may pay $20 a month. But a company using Claude for engineering, automation, and API workloads can spend six or seven figures a year.

Do more people use ChatGPT or Claude today?

Actually, more people use ChatGPT than Claude today, by a very wide margin.

The cleanest public number is OpenAI’s own 900M weekly active ChatGPT users in February 2026. Anthropic does not publish an equivalent Claude-wide weekly active user number, so we have to use third-party app estimates for the closest apples-to-apples comparison. On that basis, Sensor Tower put ChatGPT at about 1B monthly active app users in Q2 2026, versus 56M for Claude. That means ChatGPT had roughly 18x more monthly active app users than Claude.

So no, Claude is not close to ChatGPT on consumer usage. Claude is growing much faster in percentage terms, with Sensor Tower estimating 640% YoY MAU growth for Claude versus 62% for ChatGPT.

But that is growth from a much smaller base. A small product can grow 640% and still be tiny next to a product already at 1B users.

The real comparison is simple:

Metric	ChatGPT	Claude	Read
Company-reported weekly users	900M WAU	Not disclosed	OpenAI gives a clean usage number; Anthropic does not.
Third-party monthly app users	~1B MAU	~56M MAU	ChatGPT is roughly 18x larger on app usage.
Growth rate	+62% YoY MAU	+640% YoY MAU	Claude is growing faster, but from a much smaller base.
Revenue story	Huge consumer base + paid tiers + API + enterprise	Smaller user base, denser professional monetization	Claude can compete on revenue without matching ChatGPT usage.

The important point is that revenue leadership and user leadership are not the same thing.

Claude may have caught or passed OpenAI on the latest public revenue run-rate comparison, but it did not do that by becoming more widely used than ChatGPT.

It did it by monetizing a much smaller base more intensely, especially through professional, coding, API, and enterprise workflows.

ChatGPT is still the default mass-market AI product, Claude is a much smaller product with unusually high revenue density.

If you want more recent data on this point, please see our latest generative AI market report.

Does Claude make more money per user than ChatGPT?

Yes, Claude/Anthropic appears to monetize each active user much harder than OpenAI/ChatGPT, even though ChatGPT has far more users overall.

The math is brutal. ChatGPT had about 1B monthly active app users in Q2 2026, while Claude had about 56M. So ChatGPT was roughly 18x larger on app usage. But on the latest public revenue run-rate signals, OpenAI was at $20B+ ARR for 2025, while Anthropic later disclosed $30B+ run-rate in April 2026. Even with all the caveats around timing and definitions, the direction is obvious: Claude had much less consumer scale, but much more revenue density.

A rough revenue-per-app-user comparison makes the point. If we divide OpenAI’s $20B+ ARR by ChatGPT’s ~1B MAU, we get at least ~$20 per monthly active user per year. If we divide Anthropic’s $30B+ run-rate by Claude’s ~56M MAU, we get roughly ~$535 per monthly active app user per year. That implies Claude’s revenue per app user was about 25x higher than ChatGPT’s on these public numbers.

This is not a perfect ARPU calculation. OpenAI revenue includes API, enterprise, and ChatGPT subscriptions, not only app users. Anthropic revenue also includes API, enterprise, Claude Code, and cloud-platform usage, not only Claude app users. But that is exactly the point: Claude’s business is much less consumer-app dependent. Its revenue is coming from fewer, more valuable professional users and enterprise accounts.

The customer data confirms it. Anthropic had 500+ customers spending $1M+ annualized in February 2026, then 1,000+ by April 2026. Claude Code alone went from public launch in May 2025 to $500M+ run-rate by August 2025, then $2.5B+ by February 2026. Those are not normal consumer-app economics. Those are enterprise and developer-workflow economics.

So yes, Anthropic/Claude seems to monetize users better than OpenAI/ChatGPT, but we should say it precisely: Claude does not monetize the average consumer better; Anthropic monetizes its active professional and enterprise usage much more densely. ChatGPT wins on reach. Claude wins on revenue density.

Who is the author of this content?

NEW MARKET PITCH TEAM

We track new markets so founders and investors can move faster

We build living "market pitch" documents for emerging markets: AI, synthetic biology, new proteins, and more. Instead of outdated PDFs or hallucinated LLM answers, our clients get a clean, visual, always-updated view of what's really happening: key players, deals, regulations, and signals that matter. Learn more about us.