Loading prices…
STKR NewsSTKR News0 of 3 free this month
AI

The Top 10 Metrics in Anthropic’s S-1 for B2B Execs: $33 Billion in Nine Months, 900% NRR, and 12x More $100K Customers in a Year

Anthropic's latest numbers reveal a growth trajectory that defies traditional SaaS physics, forcing founders to rethink scale, retention, and the true cost of the AI arms race.

Originally on SaaStr →
AB

Adrian Boysel

Contributor

Nov 19, 2026

5 min read

Photo illustration / STKR News

We have reached a point in the AI cycle where numbers have essentially lost their meaning. When you look at Anthropic's recent S-1 filing, the figures look like typos. We are talking about $33 billion in revenue for the first nine months of the year. To put that in perspective, they grew about 1,200% compared to the same window last year. It is a level of scaling that feels less like a business and more like a force of nature.

As a builder, it is tempting to look at this and feel either inspired or completely demoralized. But if you strip away the billions, there is a blueprint here for how the next decade of B2B infrastructure is being built. Anthropic is not just selling a chatbot; they are selling the plumbing for the entire intelligence economy. For those of us in the trenches, the real value is not in the top-line revenue, but in the shifts in customer behavior buried in the metrics.

The Death of Linear Scaling

In traditional SaaS, we were taught that doubling your revenue every year was the gold standard. Anthropic just moved the goalposts into another stadium. Their revenue went from $41 million in 2023 to nearly $386 million in 2024, and then exploded into the billions. This is not linear; it is an atmospheric breakthrough.

The takeaway for founders is that AI isn't a feature—it is a replacement cycle. When companies find a model that works, they don't just add a few seats; they rewire their entire operational stack around it. This is why we see a 12x increase in customers paying over $100,000. These aren't experiments anymore. Enterprises are moving from "let's play with Claude" to "let's build our entire customer support and engineering pipeline on Claude."

The 900% Retention Anomaly

One metric in the filing stands out as particularly absurd: a 900% Net Revenue Retention (NRR). In the world of B2B software, we usually celebrate if NRR is 120% or 130%. 900% means that for every dollar a customer spent last year, they are spending nine dollars this year.

How is that possible? It comes down to usage-based scaling. In the old world, you sold seats. If a company had 1,000 employees, your ceiling was 1,000 seats. In the AI world, the ceiling is the number of tokens or API calls. As these companies integrate Anthropic's models into more automated workflows, the machine starts consuming the product, not just the humans. When machines are the customers, the traditional limits on retention and expansion evaporate.

Infrastructure as the New Moat

Builders need to pay attention to the shift toward the $100k+ customer tier. Anthropic’s growth in this segment is massive. It tells us that the "Prosumer" phase of AI—where individuals pay $20 a month for a better writing assistant—is becoming the secondary story. The primary story is the enterprise wholesale market.

If you are building an AI startup today, you have to ask yourself if you are building an app or a layer. Anthropic is winning because it has become a fundamental layer. They aren't competing with your local SEO tool; they are competing for the foundation that the SEO tool is built upon. This is a high-stakes, high-capital game, but it shows that the appetite at the top of the market is nearly bottomless for reliable, scalable intelligence.

The Cost of the Crown

We can't talk about $33 billion in revenue without talking about what it costs to generate it. While the S-1 highlights incredible growth, it also hints at the staggering capital expenditure required to keep the lights on. Training these models and serving them at this scale requires a level of compute spend that would bankrupt almost any other company on earth.

This creates a paradox for founders. On one hand, the market is ready to buy. On the other hand, the cost to compete at the frontier is rising faster than most can keep up with. Anthropic’s success is built on a massive bet that the efficiency of these models will eventually outpace the cost of the hardware needed to run them. If they are wrong, the burn rate will be legendary. If they are right, they are effectively the new utility company for the internet.

What This Means for the Builder Ecosystem

For those of us not sitting on billions of dollars in VC backing, the Anthropic numbers provide a clear signal: the middle ground is disappearing. You either provide the raw intelligence (the model) or you provide a highly specific, deeply integrated workflow that uses that intelligence to solve a problem the models can't solve on their own.

  • Stop chasing horizontal: Anthropic owns the horizontal. You cannot out-generalize them.
  • Watch the NRR: If your AI tool has low retention, you are likely just a wrapper. True utility shows up in expansion.
  • Enterprise is the prize: The 12x growth in $100k customers proves that the big money has finally stopped being afraid of the cloud-based AI stack.

The Skeptical Takeaway

Is this sustainable? Probably not in a vacuum. A 1,200% growth rate eventually hits the reality of total addressable market limits. We are currently in the "Great Integration" phase where every company is rushing to buy their piece of the future. Eventually, that rush will normalize.

However, the sheer scale of these numbers suggests that AI is not a bubble in the traditional sense. Bubbles are usually built on speculation without revenue. Anthropic has the revenue. They have the customers. They have the expansion. The question isn't whether the demand is real—it clearly is. The question is who can afford to stay in the game long enough to see the margins turn into something that looks like a traditional, healthy business.

For builders, the message is simple: the ceiling is much higher than we thought, but the floor is getting a lot more expensive. Don't try to be Anthropic. Instead, look at where they are sending those billions of dollars of intelligence and build the specialized tools that make that intelligence actually usable for the rest of the world.


Read the original at SaaStr →

The Brief

Stay Updated on Cutting-Edge Tech

A six-minute morning dispatch on the markets and the technology shaping them.

Free. No spam. Unsubscribe anytime.

Write for STKR

Become a Contributor

Earn $STKR for published stories on markets, protocols, and culture.

  • Earn $STKR for every published piece
  • Editorial support from the STKR desk
  • Byline visibility across the network
  • First look at the upcoming creator program
Apply to Write

Keep reading

All stories

Comments

24 reader responses