2 trillion US dollars, the largest IPO in the history of AI is entering the countdown.

Less than two months are left before October.

And Wall Street is waiting for a bell that may rewrite the history of technology.

If everything goes as planned, this fall, the global capital market will welcome an unprecedented behemoth — Anthropic.

$2 trillion, the countdown to the largest IPO in AI history.

Anthropic Sprinting for IPO

Dario Amodei Took Only Five Years

In June this year, SpaceX just set the record for the largest IPO with a $1.77 trillion valuation, but this myth may not even last half a year.

Quite dramatically, this $2 trillion valuation is not a price set by Anthropic itself.

The FT broke the news that six Anthropic investors each built their own models and arrived at this figure — the valuation will exceed $2 trillion when the company is listed in October.

At least one of them calculated a figure of over $3 trillion based on an 800% growth rate.

And internal executives of the company are said to have never even finalized a target price in private settings.

This company, founded by the Amodei siblings together with the former core team of OpenAI, was still crammed in a small office in San Francisco five years ago.

Today, it has been pushed to the "altar" by the spontaneous support of capital.

From 0 to $2 trillion, Anthropic took 5 years. As the benchmark of the era, Apple took a full 40 years to reach a $2 trillion market cap after going public.

14x Speed, Unstoppable Skyrocketing Growth

Anthropic's own growth curve is already unusual in itself.

At the end of 2025, its annual recurring revenue (ARR) was about $9 billion. By May this year, this figure had surged to $47 billion.

Anthropic's Q2 revenue exceeded $11.5 billion, while the figure was only $787 million in the same period last year, representing a 14-fold year-over-year increase.

Earlier the same day, Reuters exclusively revealed a previously unreported figure —

Anthropic's internal forecast shows that its revenue will reach $190 billion to $200 billion in 2028.

From $9 billion at the end of 2025, to $47 billion today, and then to $200 billion in two years.

Wall Street investment banks can no longer find a proper ruler to measure Anthropic.

Therefore, in this valuation, banks and investors have adopted the "EV/Revenue multiple" and directly used the 2028 revenue forecast.

It is not uncommon to value fast-growing software companies that are not yet profitable based on revenue instead of profit.

But skipping 2026 and 2027, and setting the pricing anchor on a financial statement that does not even exist two years later, is highly unusual.

Lines of Code Stack Up to Form a $2 Trillion Valuation

The biggest difference between Anthropic and OpenAI has always lain in their revenue structures.

70% to 80% of its revenue comes from APIs and enterprise contracts, billed by tokens. It does not rely on individual subscription users, but on the actual call volumes generated by enterprises every day with real money.

And the steepest part of this curve is Claude Code.

The ARR of this product line was only $500 million in September 2025, and had surged to $2.5 billion in February 2026. Five months, 5x growth.

According to internal company metrics, it now contributes nearly 20% of total revenue.

Data from the open source community is more intuitive: in public GitHub repositories, the proportion of commits written by Claude Code has climbed from 4% to over 10%.

Uber's CTO once stated that the company's full-year AI budget was exhausted in four months, the majority of which was spent on Claude Code and Cursor;

The average monthly API consumption per engineer ranges from $500 to $2000, and the internal adoption rate of Claude Code has risen from 32% to 84%.

Data from the latest round of enterprise surveys by Menlo Ventures shows: Anthropic accounts for about 40% of enterprise LLM spending, while OpenAI accounts for 27%; in the programming segment, Anthropic takes up about 54%, more than twice that of OpenAI (21%).

So the $2 trillion figure is actually not that mysterious when you break it down.

It is accumulated by countless calls from engineers all over the world, settled by tokens.

Prisoners Wearing Golden Handcuffs

The "Underground Resistance Network" of Genius Engineers

Recently, internal divisions within Anthropic have also been exposed.

Brian Roemmele, an early investor in Anthropic and well-known Silicon Valley whistleblower, posted: "The situation is worse than I thought. The morale of investors and employees has dropped to an all-time low."

The root cause of all this is Anthropic's senior management, a "priest class" composed of top PhDs.

Back then, when CEO Dario led a group of core members to leave OpenAI, he carried a sacred aura of "saving humanity".

Now, this aura has evolved into a "fanatical cult system".

Foreign media including Wired broke the news that the company has a unique tradition called "Dario Vision Exploration".

At the all-hands meeting every two weeks, Dario does not talk about specific business metrics, but preaches the ultimate salvation of human civilization and the destructive risks of AI in a compassionate and condescending tone.

"It doesn't feel like a company meeting, it feels like listening to a priest preach," a former employee described despairingly.

In the eyes of those arrogant executives, they are the saviors of human minds, the only prophets who know how to prevent AI from destroying the world.

Young college graduates can still be brainwashed by this grand narrative, writing code day and night for "human safety".

But as they get married, have children, and take on loans, they realize that these senior managers trapped in the "savior" illusion are becoming more and more arrogant and hostile.

At the same time, new employees do not buy into this kind of brainwashing at all.

All of this has led to internal rifts within the company.

To evade the supervision of senior management, angry frontline employees have built a hidden underground network, venting their dissatisfaction with the upper "PhD priest class" frantically.

This group of genius engineers is being torn apart painfully —

On the one hand, the company is about to usher in a $2 trillion IPO, and they hold generous stock options, so they really do not want to give up this opportunity to achieve financial freedom.

The closer the IPO is, the higher the value of the stocks in their hands, and the greater the cost of leaving.

On the other hand, the company's extreme culture is tormenting their physical and mental health. Staying may bring them hundreds of millions of dollars in wealth that changes their whole life; leaving means giving up this huge lottery ticket for spiritual freedom.

Anthropic Walking on a Tightrope, No Way Back

Anthropic right now is like a behemoth fed with massive doses of hormones, facing an unsolvable paradox.

To build the safest AI, you need the most powerful large model; to train the most powerful model, you need massive computing power; to buy computing power, you need hundreds of billions of dollars in capital.

Once capital enters the game, it will inevitably turn into a bloodthirsty beast, demanding unlimited growth and crazy returns.

Dario Amodei cannot possibly accept a $2 trillion check from Wall Street with a clear conscience, while loftily claiming that he is not subject to commercial logic.

SpaceX, which set the record in June, saw its share price plummet by 38% from its high in the following two months.

This seems to be a dangerous metaphor: the illusion of a $2 trillion valuation could shrink at any time under the heavy blow of reality.

Anthropic is walking on the thinnest and highest tightrope in the history of human technology.

Complicated government regulations, skyrocketing computing power costs, and intensifying internal governance tensions.

Any breeze in this process could be magnified into a hurricane that knocks the elephant over after the October listing.

But no matter how severe the internal rifts are, there is no doubt that Anthropic, which has the strongest coding capability across the entire network, has firmly advanced to the final round of the ASI race.

When the bell rings in October, everyone is waiting for the final outcome.

References:

https://www.reuters.com/business/anthropic-ipo-valuation-hinges-190-200-billion-2028-revenue-forecast-sources-say-2026-08-15/ 

https://www.bloomberg.com/news/articles/2026-08-14/anthropic-revenue-ahead-of-ipo-surges-over-14-fold-in-second-quarter

Edited by: Taozi Aeneas

This article is from the WeChat official account "Xinzhiyuan", Author: ASI Revelation, published by 36Kr with authorization.