Anthropic lost $80.7 billion in six months. On paper, it’s one of the largest corporate losses ever recorded, and the company plans to go public next month anyway.
The Wall Street Journal reported Tuesday that it had reviewed a draft of Anthropic’s S-1 prospectus covering the first half of 2026. The filing offers the most honest look yet at the Claude maker, and the numbers are staggering: $16.3 billion in revenue, up from $1.1 billion in the same period last year, sitting alongside a net loss of more than $80 billion. That’s about $13.5 billion lost every single month. Anthropic declined to comment on the Journal’s report.
An $80.7 billion loss, but read the fine print
Before anyone reaches for the panic button: roughly $72 billion of that loss comes from an accounting adjustment tied to Anthropic’s financing agreements, not from cash burned in operations. Paper losses from things like convertible preferred stock revaluations can swing wildly without affecting the cash the company actually spends.
The real numbers tell a more nuanced story. Anthropic spent nearly $9 billion on computing capacity and infrastructure in the first half of the year, up from $7.3 billion for all of 2025. Revenue, meanwhile, has grown at a speed rarely seen in corporate history: $4.6 billion for all of 2025, then $16.3 billion in just the first half of 2026. Last week, the company told prospective investors that third-quarter revenue alone reached $16.8 billion, up 45% from the second quarter, TechCrunch reported.
Growth like that is real. The question investors have to answer is whether it justifies what’s coming next.
The $750 billion compute tab
Buried in the filing is the number that should grab every investor’s attention: Anthropic signaled it could spend $750 billion in the coming years on computing capacity, including deals that are still being negotiated. ZeroHedge, reviewing the same document, pegged $600 billion in committed compute obligations plus $150 billion in active negotiations to reach that total.
That’s not a typo. It’s also not optional. Frontier AI companies live or die on compute, the specialized chips and data center capacity needed to train and serve models like Claude. Anthropic’s appetite for compute keeps outpacing supply, and its suppliers are charging accordingly.
In April, Anthropic signed an agreement with Amazon committing more than $100 billion over ten years for up to 5 gigawatts of additional AWS capacity. Even that hasn’t settled the matter. The company is now leasing its own data centers and working with third-party operators, a route the filing itself flags as risky since those operators have what it calls a “limited operating history,” according to TechCrunch’s reading of the document.
Your cloud provider is also your competitor
Here’s the part that makes the compute problem genuinely thorny. The S-1 describes how AWS, Google, and Microsoft, the three companies Anthropic depends on for cloud capacity, increasingly view the Claude maker as a competitor rather than a customer.
Think about that dynamic for a second. Amazon has invested billions in Anthropic. Google too. Both also sell their own AI models that compete directly with Claude. Anthropic is now asking regulators of its IPO filing to bless a business model where its most important suppliers have every incentive to squeeze it, on price, on capacity, or both.
The filing is candid about the risk. Securing enough cloud capacity is listed as one of the company’s central dependencies, and the friction with hyperscalers that now treat Anthropic as a rival makes the $750 billion spending figure look less like a plan and more like a floor.
Two customers, one-fifth of revenue
Another number investors won’t love: roughly one-fifth of Anthropic’s $16.3 billion in first-half revenue came from just two customers. The filing warns that a “significant portion” of revenue will come from a “limited number of customers for the foreseeable future.”
Customer concentration at that level is a classic IPO red flag. One churned contract can swing billions off the top line. For a company asking the public markets to pay up to $2 trillion, the valuation the Journal says Anthropic is preparing for, that’s the kind of detail that gets interrogated on earnings calls for years.
The catastrophic-risk disclosure
IPO filings always contain a risk-factors section, and they always read like corporate horror fiction. Anthropic’s goes further than most.
The draft prospectus warns that Anthropic could be held liable for damages caused by harmful AI behavior, with potential claims exceeding its available insurance coverage. It also flags uncertainty around how the company will prevent its models from causing harm, ranging from sophisticated cyberattacks to, in the filing’s own remarkable phrasing, possible blackmail by uncontrolled models.
Most S-1 risk sections read like legal boilerplate. This one reads like the company warning you that its product might be smarter than its lawyers.
Why the Anthropic IPO matters
Step back and the picture sharpens. Anthropic is attempting something nearly unprecedented: an IPO at OpenAI-adjacent scale, built on a business that spends like a sovereign wealth fund and grows like a startup. The S-1 makes both things concrete for the first time. Growth is real, and so is the burn.
The timing is deliberate. OpenAI is reportedly rushing to close a $30 billion funding round before Anthropic’s listing, with SoftBank and Abu Dhabi’s MGX in talks to join, because Anthropic’s disclosures will let investors compare the two labs directly for the first time. Whoever lists or raises second gets the anchor price set by the first.
For investors, the Anthropic IPO will be a referendum on a simple question: does frontier AI economics work, or does it just burn capital faster than anyone thought possible? The S-1 gives both camps ammunition. Revenue tripling in six months is the bull case. An $80.7 billion loss and a $750 billion compute bill is the bear case.
November will tell us which one Wall Street believes.
Sources: The Wall Street Journal (primary, reviewed the draft S-1 prospectus), TechCrunch, Bloomberg.
Frequently asked questions
How much did Anthropic lose in the first half of 2026?
Anthropic recorded a net loss of $80.7 billion in the first half of 2026, up from $13 billion in the same period of 2025, according to a draft S-1 reviewed by The Wall Street Journal. About $72 billion of that reflects an accounting adjustment tied to financing agreements, not operating cash burn.
When is the Anthropic IPO happening?
Anthropic is targeting a November 2026 public listing, according to the Journal. The IPO could value the Claude maker at around $2 trillion, though some investors have pushed for a debut valuation closer to $1.5 trillion after third-quarter revenue came in below expectations.
Why could Anthropic spend $750 billion on computing power?
The draft filing says Anthropic could spend $750 billion in coming years on cloud and computing contracts, including deals still being negotiated. Frontier AI models need enormous compute to train and serve, and Anthropic’s revenue has grown so fast that compute demand keeps outrunning supply.
Is the Anthropic IPO a good investment?
That depends on your risk tolerance. The S-1 shows explosive revenue growth, with $16.3 billion in half a year, but also extreme losses, heavy reliance on two customers for a fifth of sales, and cloud providers that increasingly see Anthropic as a competitor. This article is news, not financial advice.