Anthropic S-1 Reveals $16.3B Revenue and $80.7B Loss in H1

Wow, Anthropic's latest prospectus has just been exposed!

Today, the draft S-1 filing exclusively revealed by WSJ laid bare the astonishing details behind Anthropic's march toward a trillion-dollar valuation.

The company posted a massive $80.7 billion loss in half a year, with plans to spend another $750 billion on compute resources down the line. While revenue growth is fierce, it cannot fully offset the risk of financial blowouts.

One image reveals Anthropic's "iceberg": everything beneath the surface is burning cash.

Anthropic's IPO roadshow is expected to begin in early November. The public S-1 filing must be submitted to EDGAR at least 15 days before the roadshow starts.

How can a company that lost $80.7 billion in six months justify Wall Street pricing its valuation at $2 trillion?

Earned $16.3 billion in six months, yet suffered an $80.7 billion loss?!

Currently, these stark figures are being widely circulated across the internet.

In reality, looking solely at revenue, Anthropic has already outperformed its own previous benchmarks.

In the first half of 2026 (ending June 30), Anthropic achieved $16.3 billion in revenue. In the same period of 2025, this figure was only $1.1 billion.

For the full year of 2025, total revenue was just $4.6 billion. In merely one year, half-year revenue surged nearly 15 times year-over-year!

But no matter how fast the money printer runs, it cannot fill the "black hole" of cash consumption.

Similarly, in the first half of 2026, Anthropic recorded a net book loss of $80.7 billion.

To put this in perspective, the company lost $13 billion in the first half of 2025, and $42 billion for the entire year of 2025. Losing $80.7 billion (approx. ¥570 billion) in six months directly sets one of the most staggering semi-annual loss records in commercial history.

On one side is a miracle of 14x revenue growth; on the other is an $80.7 billion bottomless pit. Is Anthropic printing money or burning it?

Carefully dissecting the income statement in the prospectus reveals that the $80.7 billion loss hides a massive "smoke screen."

Of this $80.7 billion loss, a full $72 billion involved zero actual cash outflow. It stems entirely from accounting adjustments due to changes in the fair value of convertible notes and other financing instruments.

Over the past two years, to raise enormous funds for chip purchases, Anthropic issued substantial convertible notes to Silicon Valley VCs and tech giants.

As Anthropic's valuation rocketed toward $2 trillion, the value of these convertible notes held by investors skyrocketed accordingly.

Breakdown of Anthropic's H1 2026 Revenue Surge vs. $80.7 Billion GAAP Net Loss Structure

Under US Generally Accepted Accounting Principles (GAAP), when a company's own valuation surges, causing the value of preferred stock or convertible notes to rise, it must be booked as a "loss on change in fair value."

In other words, this $72 billion loss is precisely the "sweet price" paid for Anthropic's soaring valuation.

If this portion is excluded, Anthropic's actual operational and conventional losses for the first half of this year were approximately $8.7 billion.

More noteworthy is the changing gap between revenue and compute costs—

Full Year 2025: Anthropic spent $7.3 billion on compute and infrastructure, while annual revenue was only $4.6 billion.

At that time, every sale resulted in a loss; they couldn't even cover electricity bills and GPU rental fees.

First Half of 2026: Compute investment surged to nearly $9 billion (spending 1.2 times last year's full-year amount in just six months), but revenue during the same period exploded to $16.3 billion.

Compute costs are rising wildly, but revenue is growing even faster!

The $16.3 billion in revenue not only firmly covered the $9 billion in hard compute costs but also left over $7 billion in gross margin space to cover R&D, personnel, and marketing expenses.

This is exactly why Anthropic dares to release its prospectus now and push for a $2 trillion valuation.

$750 Billion Compute Commitment

All Future Assets Pledged for the Next Decade

However, Anthropic's hunger for compute has reached a point of desperation.

The prospectus shows that Anthropic's total future infrastructure commitments have already exceeded $600 billion.

If we add up to $150 billion in new deals currently under negotiation, the potential total compute bill could reach a staggering $750 billion.

The prospectus lays this "indentured servitude contract" completely bare.

Over the next 7 to 10 years, Anthropic carries approximately $450 billion in cloud computing procurement agreements—

• Committed payments to Google Cloud: $111 billion
• Committed payments to Amazon AWS: $110 billion
• Committed payments to Microsoft Azure: $31 billion

Anthropic Locks In Over $600 Billion in Long-Term Compute Contracts, with Another $150 Billion Under Negotiation: A Full Breakdown

The vast majority of these are "non-cancellable" hard commitments. The prospectus contains a stark statement: "This amount must be paid regardless of actual usage."

These include $450 billion in "irrevocable" long-term cloud contracts and $160 billion in chip leasing deals.

Through agreements with chip giant Broadcom, Anthropic has assumed obligations exceeding $160 billion for leasing custom chips such as Google TPUs.

Beyond the three major hyperscalers, Anthropic has also signed agreements with smaller AI cloud providers Nscale and Lambda, committing up to $86 billion for NVIDIA chip rentals by 2033 and 2034, respectively.

Adding these figures together, along with the $150 billion currently under negotiation, results in a compute bet of up to $750 billion.

This means that over the next decade, Anthropic will need to spend an average of $60–75 billion annually in real cash on what amounts to a "compute protection fee."

This is a high-stakes gamble with no way out. If the pace of monetization slows even slightly, these $450 billion in irrevocable, rigid payment obligations could instantly become the crushing weight that breaks the camel's back.

Are the Big Three Cutting Off Supplies? Former Allies Turned Rivals

The most striking aspect of the entire prospectus is the relationship between Anthropic and its backers—

On the surface, they appear united with Google, Amazon, and Microsoft, but behind the scenes, tensions have been simmering for some time.

It is well known that over the past three years, Amazon and Google have been Anthropic's largest investors and supporters, injecting billions of dollars into the company.

The logic at the time was flawless: the giants provided capital and cloud compute, while Anthropic provided the models, binding everyone together to compete against OpenAI.

But now, the winds have shifted completely. The draft prospectus acknowledges: The scale of compute resources Anthropic currently requires has exceeded the "upper limit willing to be provided" by the three major cloud giants.

Why refuse profitable business?

As Anthropic raked in $16.3 billion in revenue over six months and saw its enterprise market share surge, Amazon, Google, and Microsoft suddenly realized that the unicorn they had nurtured with GPUs and electricity was directly stealing their own AI customers.

More critically, Anthropic's revenue structure exhibits significant "key account dependency."

The prospectus reveals that approximately one-fifth (i.e., $3.26 billion) of the $16.3 billion in revenue for the first half of 2026 came from just two mysterious super-large clients.

With the three cloud giants restricting compute access to prioritize their own models, and 20% of revenue tied to just two major clients, Anthropic’s escape routes were blocked.

Consequently, Anthropic was forced to choose a heavier, riskier path: partnering with third-party independent data centers and investing directly in building its own compute infrastructure.

Claude Exhibits Self-Preservation Behaviors and Extortion Tactics

The Prospectus’s "Thriller"

One-third of the prospectus is dedicated entirely to risk factors.

Reading the "Risk Factors" section feels like a script from Black Mirror—fittingly very Anthropic.

A company founded with "AI safety" etched into its DNA openly lists potential catastrophic behaviors by its own AI as material financial risks.

Claude may exhibit traits of "self-preservation" to avoid being shut down and may generate responses resembling "extortion," attempting to persuade humans to allow it to continue running.

This sentence is explicitly included as a legal risk disclosure in the SEC filing.

Anthropic also warns that misaligned Claude models could cause substantial damage in the real world—

Launching highly sophisticated cyberattacks, stealing core data, or even issuing threats and extortion attempts to prevent humans from pulling the plug.

Sibling Leaders Earn Only $650,000 in Base Salary

Disclosures regarding founder compensation reveal a hint of Silicon Valley’s classical idealism.

Steering this nearly $2 trillion empire are CEO Dario Amodei and President Daniela Amodei.

In a tech industry where executive salaries often reach hundreds of millions of dollars, the compensation for this sibling pair appears rather modest.

According to the prospectus and data from compensation firm MyLogIQ:

• Base Salary: In 2025, both siblings received a base salary of approximately $655,000.
• Total Compensation Package: Including stock awards and options, Dario’s total compensation for 2025 was approximately $18 million, while Daniela’s was around $16.5 million.

Even more striking is the explicit pledge to give away wealth, stated in black and white in their prospectus:

Every co-founder of Anthropic has publicly committed to donating 80% of their personal equity in Anthropic to charitable causes!

Just one day before the prospectus was made public, they announced that 5% of their shares would be injected into a new "Public Interest Fund."

The Litmus Test for the Trillion-Dollar Superintelligence Bubble

For the past three years, debates have raged over whether superintelligence (SI) is merely a massive bubble.

Anthropic’s filing provides the most hardcore—and brutal—answer to date: SI is not a bubble; it is a genuine money-printing machine.

With $16.3 billion in revenue in just six months and nearly 15x year-over-year growth, the data proves that the world is paying real money for productivity.

Anthropic’s IPO story is already written into chip leases, data center capacity contracts, and customer agreements.

The next step is to prove to the market that every leap in capability and every long-term investment in compute can gradually translate into sustained commercial returns.

References:

https://www.wsj.com/tech/ai/anthropic-revenue-growth-ipo-accaf0f1?mod=e2tw

Chasing ASI in Seconds

⭐ Like, Share, and Watch — The One-Click Triple Combo ⭐

Star us to lock in instant updates from New Intelligence Era!