By Year-End, Anthropic Will Out-Earn Every Public Software Company Except Microsoft

Our jaws drop at Anthropic’s growth rate, and its valuation. But sometimes we can lose sight of just how big it already is … versus everyone else in software.

Anthropic exited 2025 at roughly $9 billion in run-rate revenue. By February 2026 it was $14 billion. Then $19 billion in March, $30 billion in April, $44 billion in early May, and $47 billion by mid-May, disclosed alongside its Series H. That is annualized run-rate, the most recent month times twelve, and Anthropic has been putting it in fundraise announcements where understating or inflating it would be securities fraud.

Hold that $47 billion up against the public software comps and the picture gets uncomfortable fast.

The only one bigger is Microsoft

Salesforce, the largest pure-play software company in the world, runs about $41 billion in revenue. Anthropic’s run-rate passed it in April. Adobe is around $25 billion. Intuit about $19 billion. ServiceNow about $14 billion. Workday about $9.5 billion. Anthropic is already larger than every one of them, three years removed from its first dollar of revenue.

On its current trajectory the run-rate is tracking toward $70 to $90 billion by December. At that level there is exactly one public software company still ahead of it: Microsoft, whose software and cloud business runs around $300 billion.

That is the headline, and it holds up, but only if you are precise about the two companies that look like exceptions.

Oracle and IBM are not the counterexamples they appear to be

Oracle just reported $67.4 billion for fiscal 2026 and guided to $90 billion for fiscal 2027. On the surface, a second company bigger than Anthropic’s year-end run-rate, for now at least. Take it apart and it isn’t. Oracle’s actual software line was $24.5 billion, and it shrank 1% last year. The growth is $34 billion of cloud revenue, most of it renting GPUs to AI labs, plus services and hardware. Oracle is now an AI-datacenter landlord with a database attached.

IBM is the same shape: a roughly $60 billion company that is mostly consulting and infrastructure, with a software segment around $30 billion.

On a software-versus-software basis, both sit below Anthropic. Microsoft is the only public company whose software business out-earns a three-year-old startup.

Anthropic will soon be bigger than all the pure-play software companies. Together.

Anthropic isn’t bigger than everyone else combined. But it’s on track to be.

Add up the pure-play public software universe and you clear $200 billion. Salesforce, Adobe, Intuit, ServiceNow, and Workday alone are over $100 billion together. Stack the rest of the field on top, sixty-plus names from Atlassian and CrowdStrike down through the long tail, and Anthropic at a $70 to $90 billion year-end run-rate is under half the total.

Take the ten next-generation software names investors watch: Palantir, Snowflake, CrowdStrike, Datadog, Zscaler, Okta, HubSpot, MongoDB, Cloudflare, and Confluent. Their combined revenue is roughly $33 billion. Anthropic’s run-rate today is bigger than all ten of them put together. Palantir, at $5.2 billion in trailing revenue, and Snowflake, at $5.0 billion, are the two largest on that list, and Anthropic has been adding more than a full Palantir to its run-rate in a matter of weeks.

That is the comparison that matters. Not one incumbent. A decade of venture-backed software winners, stacked together, smaller than a company that earned its first dollar in 2023.

It’s not classic “ARR”, but who cares

Two caveats, because the post is worthless if a CFO can take it apart in the comments.

First, run-rate is not annual revenue. The $47 billion is one strong month annualized. The public comps above are trailing-twelve-month actuals. On actual calendar-2026 revenue, Anthropic will land somewhere around $20 to $26 billion for the year, which puts it in the top five of software, not at number two. The year-end run-rate tells you where the line is heading. The trailing revenue is where it has been. Both are real, and the gap between them is the entire point of an exponential.

Second, Anthropic books a meaningful share of revenue gross. Sales through AWS, Google, and Microsoft are counted as full end-customer spend, with the partner’s cut booked as a cost. That inflates the top line relative to a net-reporting B2B company. It does not change the trajectory, but it means the run-rate and a clean subscription number are not the same animal.

Neither caveat dents the core finding. They keep it defensible.

It’s why many VCs struggle to get excited about a lot of B2B software now

The reason this matters is not bragging rights for one AI lab. It is what the comparison says about how software gets paid for.

Anthropic did not get here selling seats. It got here selling tokens, and a single developer running an agent against a real codebase consumes at a rate no per-seat plan ever modeled. Claude Code alone went from zero to $2.5 billion in run-rate in about nine months, larger than most of the companies on that chart, built by one product team and priced entirely by consumption.

The public software names sitting below Anthropic are, with few exceptions, seat-based businesses. The market has already started repricing them, with the public software multiple at decade-plus lows. The revenue gap is the tell. When the fastest-scaling software company in history monetizes work instead of logins, every roadmap built on counting users is now competing with a model that gets paid for output.

The question for founders is not whether Anthropic’s run-rate is real. It is whether your pricing survives in a market where the biggest new software company on earth never sold a seat.