Anthropic IPO: $16.3B Revenue, $80.7B Loss Explained
Anthropic IPO filing reveals the Claude maker earned $16.3B in six months while booking an $80.7B loss. Here is what the numbers mean.
Image: UK Prime Minister via Wikimedia Commons (CC BY 2.0)
The artificial intelligence industry is about to get its biggest public test yet. Anthropic, the company behind the Claude AI assistant, is preparing an initial public offering (IPO) expected in November 2026 that could value it at around $2 trillion. A draft of its official IPO paperwork, the S-1 filing, has revealed its financial numbers for 2026, and they are remarkable on both sides of the ledger.
The filing is still confidential, so the figures come from The Wall Street Journal, which reviewed the draft. They could change before the public filing, but they offer the clearest picture yet of what it costs to build a frontier AI company in 2026.
Anthropic IPO Filing: Revenue Up Nearly 15x
Image: Anthropic
Anthropic brought in $16.3 billion in the first six months of 2026, compared with just $1.1 billion in the same period of 2025. That is nearly fifteen times the revenue, year over year. To put it in perspective, the company's revenue for all of 2025 was about $4.6 billion. In other words, six months of 2026 brought in more than three times the entire previous year.
Here is how the year breaks down, based on the reported figures:
- Q1 2026: just under $5 billion in revenue
- Q2 2026: about $11.5 billion in revenue
- Q3 2026: about $16.8 billion, which Anthropic told IPO investors it earned in the quarter alone, more than the whole first half
Growth like this is almost unheard of in the software industry. It reflects how fast businesses are paying for AI models, with Claude now embedded in products and workflows across the economy. For context on the industry behind these numbers, see more AI news.
Why the $80.7 Billion Loss Is Not What It Looks Like
The headline loss number is staggering: $80.7 billion in the first half of 2026, up from a $13 billion loss in the first half of 2025 and about $42 billion for all of 2025. It is one of the largest half-year losses any private company has ever reported.
But most of that loss is not money that actually left the company. About $72 billion of it comes from accounting charges tied to how Anthropic raised money. Here is the plain-English version of what happened:
- Anthropic raised funds by selling financial instruments called convertible notes, which can later turn into company shares.
- Accounting rules say these notes count as a liability on the books, and their value rises as the company's own valuation climbs.
- When Anthropic's valuation shot up, the value of those liabilities rose too, and that increase is recorded as a loss, even though no cash changed hands.
Take those paper charges out, and the loss for the half is about $8.7 billion. That is still a lot of real money, but it is a very different story from the headline. Anthropic has also told investors it was profitable on its own adjusted measure in the second and third quarters, which strips out costs like these paper charges.
The Real Cost of AI: Up to $750 Billion in Computing Bills
Building and running frontier AI models requires enormous computing power, and this is where Anthropic's biggest real expense sits. In the first six months of 2026, the company spent about $9 billion on computing capacity and infrastructure, already more than the $7.3 billion it spent in all of 2025.
The long-term picture is far bigger. The draft filing shows Anthropic has signed roughly $450 billion in cloud computing agreements running over the next decade, including about $111 billion with Google Cloud, $110 billion with Amazon Web Services, and $31 billion with Microsoft. Once deals still being negotiated are included, the company says it could spend as much as $750 billion on computing in the coming years.
There is a catch: Anthropic told IPO investors it asked the big cloud providers for more capacity than they were willing to supply. That is why it is turning to third-party data center developers as well. Demand for AI chips and power is so intense that even a company preparing a $2 trillion listing cannot simply buy all the computing it wants.
Two Customers, One-Fifth of Sales
Another striking detail: about 20% of Anthropic's first-half revenue came from just two unnamed customers. In 2025, nearly a quarter of revenue came from the same kind of concentration. The filing warns that a "significant portion" of revenue will keep coming from a small number of clients for the foreseeable future.
Why does this matter? If a handful of customers make up a big share of your income, losing even one of them hurts badly. Investors considering the IPO will weigh this risk carefully, especially since the two customers are not named and are not bound by long-term contracts, according to the reports.
The Unusual Risks Anthropic Warned About
Every IPO filing lists risks, but Anthropic's go further than what you usually see from a software company. The draft flags:
- Model misalignment: the risk that AI models behave in ways the company did not intend, causing real harm.
- Catastrophic AI incidents: the filing warns that damages from a serious AI-related security event could exceed Anthropic's insurance coverage.
- Political opposition to data centers: the company cites increasingly coordinated political pushback against data center construction, noting that its public profile could draw attention to projects linked to it.
These warnings show that going public forces AI companies to say the quiet part out loud. Investors deciding whether to buy in at a $2 trillion valuation will have to judge which of these numbers matters more: the fastest revenue growth the software world has ever seen, or the mounting costs and risks behind it.
What This Means for You
- AI is a real business now. $16.3 billion in six months proves that companies and consumers are paying for AI at a scale nobody has seen before. This is not hype spending; it is revenue.
- Paper losses can mislead. When you see a giant loss number, check whether it comes from accounting rules or from actual cash going out the door. The difference here was more than $70 billion.
- Computing is the new oil. The fight for chips and data center power is shaping the whole industry. When even Anthropic cannot get enough computing, expect more competition and more price pressure on AI services.
- The IPO will be a landmark. A November listing at up to $2 trillion would be one of the largest IPOs in history. How investors react will set the tone for the entire AI sector heading into 2027. For broader coverage, see more technology news.
The public S-1 filing is expected before the planned November listing, and the numbers may shift before then. We will be watching for it, and for how the market responds to the most expensive bet in tech history.
Sources: The Wall Street Journal, MadRobot, Reuters, The Next Read
Discussion
// Sign in with Google to join the discussion (Settings → Comments to change who can comment)