Anthropic Only Has ~5,000 Employees. Almost No One Has Ever Been This Efficient. That’s By Choice.

Anthropic just hit $30 billion in annualized revenue.

$30 billion. In a company founded in 2021. With somewhere between 3,000 and 5,000 employees, depending on which source you trust.

That is the not just the fastest $30 billion revenue run in the history of technology. It’s the leanest. And it’s not even close.

The Revenue Ramp That Broke Every Record

Brad Gerstner at Altimeter (an investor in both Anthropic and OpenAI) laid out the revenue trajectory on a recent podcast, and the numbers are staggering:

End of 2024: $1B annualized run rate Mid 2025: $4B End of 2025: $9B End of March 2026: $30B

That’s 30x growth in 15 months. Anthropic hit their year-end 2026 revenue target by Q1. To put the monthly adds in context, Gerstner noted Anthropic added the equivalent of Databricks plus Palantir combined in revenue in a single month.

He wouldn’t be shocked if Anthropic exits 2026 at $80-100 billion in revenue.

No technology company has ever done this. Not Google. Not Meta. Not Microsoft. Not Salesforce. Not even OpenAI.

The Real Story: Revenue Per Employee

But the revenue number isn’t the most remarkable part. The headcount is.

Estimates on Anthropic’s current employee count vary. Tracxn pegs it at ~5,000 as of March 2026. TrueUp says ~3,000. PitchBook and Brad Gerstner both referenced ~2,500 earlier this year. Even at the high end, the revenue-per-employee ratio is something we’ve never seen before in enterprise technology.

Here’s how every major tech company looked when they crossed the $30 billion annual revenue threshold:

Anthropic is generating 6-10x more revenue per employee than Google did at the same scale. And 15-25x more than Salesforce. And OpenAI, while adding headcount faster, isn’t far behind.

Google crossed $30 billion with 32,000 people. Salesforce needed 79,000. Anthropic did it with somewhere around 5,000 at most.

Focus as a Competitive Weapon. And Letting AI Handle Everything the Best Humans Can’t.

This isn’t because they haven’t had time to hire. It’s by choice.

No multimodal sprawl. No video. No hardware. No chips. No building data centers. They picked coding and enterprise, went deep, and executed with a fraction of the people.

Gerstner credited Anthropic’s discipline in saying no to everything except the core bet. That focus, combined with a clear coding lead, is what took them from “counted out” a year ago to running ahead of OpenAI on revenue.

The biggest cost input for an AI company is compute, and Anthropic only has ~1.5-2 GW of capacity. That compute cost is relatively fixed whether revenue is $1 billion or $80 billion. So gross margins are expanding as revenue scales up against a somewhat fixed cost base. Inference costs are down 90% year over year.

OpenAI is Just as Efficient. At $24B Run-Rate It Has 4,500 Employees (and Plans to Double)

OpenAI is currently at about $24 billion in annualized revenue with roughly 4,500 employees. They just announced plans to nearly double headcount to 8,000 by end of 2026.

It’s essentially as efficient today as Anthropic from a headcount perspective, but even if it doubles, it will still be far more efficient than any other traditional software company.

Lean in the Age of AI Is a Choice. And One That Seems to Be Able to Scale … Almost Indefinitely.

Three takeaways for founders and operators:

1. The “lean AI company” isn’t a temporary phase. It’s the new model.

Anthropic isn’t lean because it’s early. It’s lean because AI companies don’t need 79,000 employees to hit $30 billion. The ratio of intelligence output to human headcount has permanently shifted. Every B2B company should be asking: what does our org chart look like if we take this seriously?

2. Focus beats portfolio.

Anthropic said no to almost everything. They picked coding and enterprise co-work. They went deep instead of wide. In a market moving this fast, the companies that win are the ones that concentrate bets, not spread them.

3. Revenue per employee isn’t everything. Growth still trumps is. But it’s a defining metric of the AI era.

Forget revenue per dollar of funding. The companies defining the next decade of B2B will be the ones generating $5M, $10M, or more in revenue per employee. If you’re building a company in 2026 and your revenue-per-employee trajectory looks like Salesforce in 2005, you’re building the wrong kind of company.

Anthropic is not a 5,000-person company because it hasn’t had time to hire. It’s a 5,000-person company because that’s all it needs. That’s the most important signal in B2B right now.

If your company’s ratio looks like Salesforce in 2005, as epic as it was for its day… you’re building the wrong kind of company.

And it’s not just because the best in AI can be leaner. It’s because they want to be.

Wouldn’t you, if you could? Just the best of the best? And let AI handle the rest? Of course you would.

Can You Do It, Too? SaaStr Has. More Will, Too. For Us, It’s Now a Choice.

We’re not Anthropic. We’re not building frontier AI models. But we are running SaaStr — an eight-figure B2B media, events, and AI platform — with 3 humans and 20+ AI agents. We went from 20+ employees and a second office to a team you can count on one hand. Revenue swung from -19% to +47% year over year. We produce triple the content. Our AI agents have sourced millions in pipeline and closed real deals on their own.

It’s more work for the 3 humans, not less. But it’s better work. And at this point, we aren’t going back. Not because we can’t afford to hire. Because we don’t want to.

That’s the part people miss about the Anthropic story. It’s not just efficiency. It’s preference. Once you’ve operated at this ratio … once you’ve seen what a small, focused team can do with AI handling the rest … adding headcount feels like adding friction. You just move too fast now to wait for the mediocre to get things done. No time anymore. Or patience. It’s easier to manage AI Agents to do it. It really is.

More companies will get here. Faster than most people think.