To the New Humans of the New World — “Massive Losses and Profits at the Same Time”
Revenue of $16.286 billion in the first half. A net loss of $80.7 billion.
Those two figures appear in the same prospectus.
On October 26, Anthropic filed an S-1 with the SEC under the ticker ANT, proposing to list on Nasdaq. The market’s first reaction was excitement; the second was confusion: how can one company be both a growth machine and the AI company with the biggest loss on paper?
The answer lies in the accounting.
Of the $80.7 billion loss, about $72.4 billion came from changes in the fair value of convertible notes, derivative liabilities, and SAFEs. The faster the valuation rises, the larger the revaluation losses on these instruments become on the books. This isn’t cash that was burned.
But that doesn’t mean the company is already profitable.
Operating losses in the first half were still $693 million, with another roughly $8.76 billion in other expenses. According to Reuters, at an investor day the company said third-quarter revenue was about $16.8 billion and that it was profitable on an adjusted basis. That was a verbal statement, not an audited figure in the prospectus. The S-1 itself is explicit: the company may not achieve or sustain profitability.
What’s really worth watching is the growth curve.
$41 million in 2023, $386 million in 2024, $4.585 billion in 2025, and $16.286 billion in the first half of 2026. It has more than 11,900 customers spending over $100,000 a year. That curve is real.
The question is: what’s supporting that growth?
According to Reuters, in 2025 about 47% of sales came through Amazon and Google’s cloud channels. Reuters also reported that two unnamed customers each accounted for about 12% of revenue in 2025. The channels, landlords, shareholders, and competitors are often the same group of companies.
Dependence is a cost in its own right.
An even greater cost is computing power.
The S-1 shows that as of June 30, fixed or minimum commitments totaled about $265.6 billion, leases that had not yet commenced totaled about $321.4 billion, and roughly $245 billion in new or amended commitments had been signed after June. When revenue is growing rapidly, this is a moat. If revenue slows, it becomes a rigid expense.
The same commitment means two different things in two different cycles.
Then there’s control.
Seven co-founders hold 50.1% of the voting power through a single Class F share, while long-term benefit trusts can elect a majority of the board. People who buy the stock get the right to returns, not the steering wheel.
That isn’t necessarily a bad thing. But you need to understand what you’re buying.
As for the reported $100 billion fundraising target and roughly $2 trillion valuation, those are currently just media reports citing people familiar with the matter. The S-1 has no price range yet.
Without a price, you can’t call it cheap—or expensive.
A good company, a good stock, and a good price are three different things.
Anthropic has shown that the first may be possible. The other two will have to wait for the price range, official quarterly results, and lockup terms to provide answers.
Until those answers arrive, watching is itself a judgment.
The above is for sharing information and views only and does not constitute investment advice.
Half-year revenues soar to billions,
Yet on the books, losses fill the page.
Compute commitments tower like mountains;
Without a price, don’t rush onstage.
— Mingdao
Revenue of $16.286 billion in the first half. A net loss of $80.7 billion.
Those two figures appear in the same prospectus.
On October 26, Anthropic filed an S-1 with the SEC under the ticker ANT, proposing to list on Nasdaq. The market’s first reaction was excitement; the second was confusion: how can one company be both a growth machine and the AI company with the biggest loss on paper?
The answer lies in the accounting.
Of the $80.7 billion loss, about $72.4 billion came from changes in the fair value of convertible notes, derivative liabilities, and SAFEs. The faster the valuation rises, the larger the revaluation losses on these instruments become on the books. This isn’t cash that was burned.
But that doesn’t mean the company is already profitable.
Operating losses in the first half were still $693 million, with another roughly $8.76 billion in other expenses. According to Reuters, at an investor day the company said third-quarter revenue was about $16.8 billion and that it was profitable on an adjusted basis. That was a verbal statement, not an audited figure in the prospectus. The S-1 itself is explicit: the company may not achieve or sustain profitability.
What’s really worth watching is the growth curve.
$41 million in 2023, $386 million in 2024, $4.585 billion in 2025, and $16.286 billion in the first half of 2026. It has more than 11,900 customers spending over $100,000 a year. That curve is real.
The question is: what’s supporting that growth?
According to Reuters, in 2025 about 47% of sales came through Amazon and Google’s cloud channels. Reuters also reported that two unnamed customers each accounted for about 12% of revenue in 2025. The channels, landlords, shareholders, and competitors are often the same group of companies.
Dependence is a cost in its own right.
An even greater cost is computing power.
The S-1 shows that as of June 30, fixed or minimum commitments totaled about $265.6 billion, leases that had not yet commenced totaled about $321.4 billion, and roughly $245 billion in new or amended commitments had been signed after June. When revenue is growing rapidly, this is a moat. If revenue slows, it becomes a rigid expense.
The same commitment means two different things in two different cycles.
Then there’s control.
Seven co-founders hold 50.1% of the voting power through a single Class F share, while long-term benefit trusts can elect a majority of the board. People who buy the stock get the right to returns, not the steering wheel.
That isn’t necessarily a bad thing. But you need to understand what you’re buying.
As for the reported $100 billion fundraising target and roughly $2 trillion valuation, those are currently just media reports citing people familiar with the matter. The S-1 has no price range yet.
Without a price, you can’t call it cheap—or expensive.
A good company, a good stock, and a good price are three different things.
Anthropic has shown that the first may be possible. The other two will have to wait for the price range, official quarterly results, and lockup terms to provide answers.
Until those answers arrive, watching is itself a judgment.
The above is for sharing information and views only and does not constitute investment advice.
Half-year revenues soar to billions,
Yet on the books, losses fill the page.
Compute commitments tower like mountains;
Without a price, don’t rush onstage.
— Mingdao