In the nearly six years since a handful of ex-OpenAI employees founded rival AI lab Anthropic (ANTH.PVT), the company's revenue has climbed to $17 billion in the last quarter alone, jumping 12x year over year.
For the nine months that ended Sept. 30, Anthropic's sales eclipsed $33 billion. It brought in $2.5 billion in all of 2025 and just $41 million in 2023.
CEO Dario Amodei has quickly become one of the key figures of the AI explosion alongside rival OpenAI (OPAI.PVT) CEO Sam Altman.
Anthropic has also drawn its share of controversy ranging from employees predicting that AI could kill all humans to showdowns with the Trump administration.
Despite that, Anthropic's sales continue to improve. Its number of customers spending more than $100,000 has increased from roughly 1,500 in December 2025 to more than 11,900 as of Sept. 30. Its computing efficiency has also jumped, meaning it's seeing more revenue per dollar spent on computing infrastructure.
"There's no real historical analogy for a company that has been this successful this fast," NYU Stern School of Business professor Arun Sundararajan told Yahoo Finance.
And with Anthropic set to launch its initial public offering late next week, it's easy to see why it's Yahoo Finance's Company of the Year for 2026 during its annual Invest Conference in New York City.
Anthropic was founded in 2021 by Amodei, his sister Daniela Amodei, Tom Brown, Jack Clark, Jared Kaplan, Sam McCandlish, and Chris Olah. Each of them worked in various capacities at OpenAI.
Menlo Ventures partner Matt Murphy, who led the company's investment in Anthropic, said he was attracted to the idea that Dario Amodei and his cofounders stuck together after leaving OpenAI to create their own lab.
"Amodei brought over this great team of founders from OpenAI, so you kind of felt like, 'Hey, this is a band that wanted to get back together', " Murphy explained.
"That reduces some risk in any startup when a group of founders really … wanted to work together and had had work experience with each other and kind of trust with each other," he added.
Amodei was already a figure in AI circles before founding Anthropic, but became a household name as a result of his outspoken calls for AI safety and warnings that the technology could lead to job losses or worse.
That focus on AI safety is part of the reason Amodei left OpenAI to co-found Anthropic.
In its S-1 investor prospectus, Anthropic lists the myriad ways its AI could cause harm if not properly monitored. And in September, Amodei penned an essay calling for AI labs to slow the pace of frontier AI development.
Anthropic's success comes down to a combination of its powerful AI models, its focus on the enterprise, and its Claude Code platform in particular.
Claude Code, an agentic coding tool, was a catalyst for the company's growth. That's since expanded to Anthropic's Claude Cowork, which Anthropic says customers are adopting faster than they did when Claude Code was initially released.
"They made an early bet on focusing on the enterprise as the primary use case, and really doubled down on that in terms of … coding, and really moved further along in that direction than OpenAI," explained Santa Clara University Leavey School of Business professor Ram Bala.
"That paid off for them incredibly well."
While Anthropic's growth has been explosive, the company also faces headwinds.
The midterm election results, in which Democrats took control of the House and Senate, have been interpreted, in part, as pushback against the AI data center build-out and AI generally.
That's raised concerns about whether data center construction will be slowed, potentially impacting future AI model development.
Anthropic is also still contending with the fallout from the Trump administration's decision to label it a supply chain risk after the company said it wouldn't allow the Department of Defense to use its models to develop autonomous weapons or for the mass surveillance of Americans.
The competitive landscape is also as fluid as ever.
OpenAI has dramatically expanded its enterprise offerings as it seeks to better compete with the company. Google (GOOG, GOOGL) and Meta (META) are also gaining ground in overall model capabilities and expanding their enterprise platforms.
"The revenues are eye-popping in terms of the growth," said University of Pennsylvania Wharton School professor David Hsu.
"I think it does remain to be seen the durability of that in terms of how far these scaling laws go and into what other domains beyond what we know them to be excellent at now. It's not like a mature business. Anthropic is only five years old itself," he added.
Anthropic, like its closest rivals, also has to contend with changing market dynamics. The age of so-called "tokenmaxxing," during which companies allowed workers to burn through their AI allotment to find the best use cases, is over.
Instead, enterprise companies are increasingly looking for ways to save on AI, opting to use less expensive models from Anthropic and its ilk, or cheaper open-weight models depending on their needs.
"I think … usage from Anthropic models is going to be at the frontier. It's going to be for complex use cases, scientific discovery, cybersecurity, and other kinds of tasks. And 80% of the usage of models is going to be pretty much open weights," Bala said.
According to the model-picking platform OpenRouter, open-weight models accounted for 8 of the top 10 models by…
