This is unbelievable! Anthropic has skyrocketed, capturing 70% of the US market share and leaving OpenAI with only 30% of the market. Even more outrageous is its annualized revenue, which is nearly $20 billion, a surge of $5 billion in just two weeks. Now, everyone is voting with their feet against Claude.
Anthropic not only won, but won decisively!
This image is now trending across the entire internet—
In February 2026, Anthropic's market share in the United States surged by nearly 70% , quickly overtaking OpenAI.
In just one year, ChatGPT's original 90% market share was largely taken over by Claude .
Even more impressively, Anthropic's annualized revenue (ARR) has once again broken records, approaching $20 billion .
It surged by $5 billion in just two weeks!
With the median revenue of $16.64 billion among the Fortune 500 companies as a reference, one has to marvel at how enormous Anthropic's scale is.
Annual income doubled, soaring by 20 billion US dollars.
Anthropic's revenue performance is truly astonishing, especially considering that it has become the "number one company" blocked by government agencies across the United States.
It can be said that it is caught in an extreme situation of "half seawater and half fire".
On the one hand, Anthropic's ability to raise money is astonishing.
Dario Amodei revealed at a conference that the company's annual revenue has soared to nearly $20 billion.
Compared to $9 billion at the end of last year, this has more than doubled . Looking at a longer time frame, the rate of revenue growth is truly astonishing.
This phenomenal growth is largely attributed to Claude's strong performance.
In particular, starting from the beginning of 2026, Claude Code unexpectedly became a hit across the internet, and the flagship Opus 4.6 completely transformed Claude.
In February, Claude Cowork emerged and updated more than a dozen plugins, causing a global software stock crash and wiping out nearly a trillion dollars in market value.
Last month, Anthropic also announced a new round of financing of $30 billion, bringing its valuation to a peak of $380 billion.
At the same time, an impressive report card was released, sending shivers down the spines of the entire Silicon Valley—
Its run-rate revenue has soared to $14 billion, achieving explosive growth of more than 10 times per year for three consecutive years .
The more dominant metric is enterprise penetration rate.
Over the past year, the number of customers spending more than $100,000 annually on Claude has increased sevenfold ; the number of major customers with annual spending exceeding $1 million has surged from a dozen two years ago to more than 500.
Eight of the Fortune 10 companies have become loyal Claude users.
Claude Code alone, which only launched its agent programming tool last May, has already generated over $2.5 billion in annual revenue.
With its smooth programming experience, Claude Code has garnered a huge reputation among developers.
A startling analysis reveals that 4% of publicly available code commits on GitHub worldwide are automatically generated by Claude Code, a figure that doubled in just one month.
Today, Claude Code has become an essential AI tool for many companies around the world, deeply integrated into their workflows.
Giants including Shopify, NASA, Figma, and Stripe are lining up to send money to Anthropic.
Total Victory
As netizen Deedy put it, Anthropic is achieving a "complete victory".
According to the latest statistics from Similarweb, Claude.ai is experiencing a surge in traffic in the second half of February, far exceeding that of Grok and DeepSeek.
The latest Ramp AI index shows that Anthropic's enterprise coverage surged from 16.7% to 19.5%, while OpenAI's declined to 35.9%.
One in five companies now pays for Anthropic. A year ago, the ratio was 25:1.
In terms of API access spending, Anthropic holds a dominant 90% market share, making it virtually invincible.
Moreover, while confronting the Pentagon, Anthropic also gained a huge influx of traffic.
Claude's app is far ahead of the others on the App Store download charts in Google Play, the US, Canada, France, and other regions.
Claude even experienced multiple global outages due to factors such as an excessive number of users.
However, amidst this capital frenzy, Anthropic kicked a hornet's nest .
Because of its insistence on "no compromise" on AI safety principles, it directly clashed fiercely with the Pentagon, ultimately leading Trump to blacklist it.
U.S. Defense Secretary Pete Hegseth also characterized it as a "supply chain risk" and cut off government business ties with it.
The shadow cast over Anthropic's business prospects has also affected many other related companies.
The Pentagon split has resulted in the birth of its first "victim".
According to an exclusive report by Information, Anthropic's collaboration with Palantir may become the "first casualty" of this dispute .
For more than a year, Anthropic has been providing services to the U.S. government through the software giant Palantir.
Palantir's system incorporates the Claude model in a deeply integrated manner.
The Pentagon and various federal agencies use Claude to find patterns in massive amounts of classified data to aid in critical decision-making.
However, this major decision by the Ministry of National Defense directly brought this "marriage" to an end—
The U.S. government has explicitly stipulated that contractors are restricted from using any of Anthropic's technologies.
For Palantir, whose revenue from government contracts accounted for 42% of last year's total, this is undoubtedly a "must-choose" question.
Although Anthropic's revenue is very impressive, losing the endorsement of a top government contractor like Palantir is a heavy blow to its market position.
The Pentagon has been using Palantir software in conjunction with Claude, which is managed by AWS.
Sources familiar with the matter revealed that Palantir has prepared a backup plan.
With just a few weeks of fine-tuning, the integrated Claude can be replaced with models from OpenAI and Google, without affecting contract revenue.
On the other hand, a full replacement of Claude will also face a six-month "growing pains" transition period.
At a recent defense technology summit, Palantir CEO Alex Karp made sharp remarks, criticizing Silicon Valley for standing in opposition to the U.S. military.
He warned that if Silicon Valley is both vying for white-collar jobs and stabbing the military in the back, it will only lead to the nationalization of technology.
Karp emphasized, "This is the ultimate fate of this path!"
Palantir currently allows customers to choose AI models from different vendors such as Anthropic, OpenAI, and Google when analyzing data on its platform.
Previously, an Axios report stated that restrictions imposed by the US government could put Anthropic at risk of being cut off from supplies by Nvidia.
In theory, it is also possible to sever its ties with cloud service providers such as Amazon and Google, which also hold defense contracts.
In this game, OpenAI quickly rose to become the substitute, while Ultraman handed over GPT, much to the disappointment of the entire Silicon Valley.
OpenAI urgently revises its agreement.
Ultraman frankly admits: His eating habits are unsightly.
Just today, a lengthy internal post from Ultraman revealed further details of their cooperation with the Ministry of Defense.
The core message is that OpenAI urgently patched the agreement by adding a series of legally binding clauses.
He also emphasized that the ban covers all personal privacy data obtained through commercial means, in order to prevent AI from becoming a tool for spying on the public.
Interestingly, Altman clearly drew a line: intelligence agencies such as the National Security Agency (NSA) are currently prohibited from using OpenAI's services.
If these giants want to access GPT, they must go through the cumbersome contract modification process all over again.
He also admitted that hastily announcing the collaboration last Friday was a mistake.
Faced with such a high-risk decision, this eagerness to "climb the ladder" is too ugly and has shocked the entire internet.
In fact, a heated debate about soul and contract also broke out within OpenAI.
On Tuesday, at an all-hands meeting, Altman responded to employee questions about the Pentagon order in an unprecedented manner.
He stated bluntly that OpenAI cannot determine how the Department of Defense specifically uses its technology.
The final action button was in the hands of Defense Minister Pete Hegseth.
So this means that OpenAI's stated "principles" are just for show, meant to impress its own employees and the general public.
ChatGPT crashed! 2.5 million people unsubscribed.
Ultraman tried to turn the tide, but he couldn't stop the ChatGPT army from going berserk.
Today, the "Boycott ChatGPT" movement, which has swept the internet, has garnered 2.5 million members worldwide .
On Reddit, unsubscribing from ChatGPT has become a "politically correct" choice.
Recently, OpenAI announced a new round of financing of $110 billion, claiming that it will have 900 million weekly active users by the end of 2025, but QuitGPT's momentum should not be underestimated.
On Instagram, related posts have been viewed over 36 million times, and more than 17,000 people have signed a boycott agreement on the official website.
New York University marketing professor Scott Galloway bluntly stated that only by "voting with one's wallet" can these tech giants, who are deeply intertwined with political power, feel pain.
A staggering $19 billion in growth and 2.5 million people canceling their subscriptions—the AI industry has never been so polarized.
Behind Anthropic's "total victory" lies a break with the old order; while OpenAI's "emergency patch" is a helpless response to the laws of survival.
This epic battle of power, money, and principles is only just beginning to reach its climax.
When the principles of technology clash with the ironclad power, who will have the last laugh?
References:
https://x.com/Yuchenj_UW/status/2028974344710606905
https://www.bloomberg.com/news/articles/2026-03-03/anthropic-nears-20-billion-revenue-run-rate-amid-pentagon-feud?embedded-checkout=true
https://www.reddit.com/r/technology/comments/1rjxe9f/15_million_users_leave_chatgpt/
https://x.com/aaronpholmes/status/2028942999548297464?s=20
This article is from the WeChat official account "New Zhiyuan" , author: New Zhiyuan, editor: Peach, and published with authorization from 36Kr.
