I spent the last 24 hours reading Anthropic’s brand new 400+ page S-1.

An S-1 is the massive document companies file with the SEC before going public.

It contains everything from financials and risk factors to executive compensation and business strategy.

Anthropic’s S-1 is particularly interesting because we’ve never gotten a detailed look inside one of the world’s leading AI companies.

And some of the numbers are absolutely insane.

So instead of making you read hundreds of pages of legal jargon, I pulled out everything you really need to know.

Including some of the weird stuff.

Let’s get into it.

What Is Anthropic?

Anthropic is the company behind Claude, one of the biggest competitors to OpenAI’s ChatGPT.

Dario Amodei is the CEO and co-founder.

He previously worked at OpenAI, where he was vice president of research.

But in late 2020, Dario and several other researchers left following disagreements over the direction of the company.

The major issue was how quickly AI should be developed and how much emphasis should be placed on safety.

So in 2021, Dario and six other former OpenAI employees founded Anthropic.

Their goal was to build extremely powerful AI while putting safety at the center of the company.

That philosophy still runs through Anthropic today.

In fact, Anthropic isn’t technically a normal corporation.

It’s a public benefit corporation with the stated purpose of responsibly developing advanced AI for the “long-term benefit of humanity.”

In a letter attached to the S-1, Anthropic even stated:

We encourage you, as a prospective owner of our stock, not to think of your purchase in purely transactional terms. We hope you will see it instead as an opportunity to be part of something greater… Our aim is to help humankind chart a safe course through this technological evolution… Everyone who owns our stock, whether an institution or an individual, becomes part of that effort. We hope you will share our ambition to see it through.

Does Anthropic Make Money?

Absolutely. This is one of the fastest growing companies of all-time.

Here’s Anthropic’s revenue by year:

• 2023: $41 million

• 2024: $386 million

• 2025: $4.6 billion

• First six months of 2026: $16.3 billion

Anthropic generated nearly four times more revenue in the first six months of this year than it did during all of 2025.

Plus, the company separately reported its annualized revenue run rate hit $65 billion at the end of July 2026.

For some perspective, Google grew its revenue 11X in two years before its IPO.

Anthropic increased its revenue 112X in the same amount of time.

And Anthropic is only accelerating from here.

Yeah… But Is Anthropic Profitable?

Not yet. This is the knock on the company.

Anthropic reported an $80.7 billion net loss during the first six months of 2026 on $16.3 billion in revenue.

That sounds horrifying. And many media outlets are focusing on this number.

But Anthropic didn’t actually burn through $80 billion in six months.

The majority of that loss was an accounting oddity.

As the company's valuation soared, accounting rules required Anthropic to record a $72.4 billion loss on investments that could eventually convert into Anthropic stock.

In other words… the bigger Anthropic gets, the greater this loss on investments.

The rule makes them a victim of their own success.

The real number to look at is operating loss – which totaled $693 million in the first six months of 2026.

Still a loss.

But a much different story.

What Comes Next for AI?

Anthropic believes AI will become as smart as the world's best doctors, scientists, engineers and programmers.

And with millions of these AIs capable of working around the clock, Anthropic describes the future as “a country of geniuses in a data center.”

So what would millions of geniuses actually do?

• Thousands of AI scientists could work simultaneously on cancer treatments, new drugs and personalized medicine.

• AI engineers could design better batteries, power grids, computer chips and manufacturing processes.

• AI programmers could write and test enormous amounts of software around the clock.

• AI tutors could provide personalized education to anyone with an internet connection, while AI doctors, lawyers and financial experts could make specialized knowledge available to billions more people.

• Anthropic says AI could even bring developing countries “capabilities that took wealthy nations generations to build.”

And then there's AI itself.

Claude is already helping Anthropic's researchers run experiments, process training data, write code and build future versions of Claude.

So millions of AI researchers could eventually work on making AI itself smarter.

That's how you get from today's chatbots to Anthropic's claim that AI could transform the economy in ways “greater than the impact of industrialization, electricity, semiconductors, and the internet.”

What Are the Risks?

This is the strangest risk section I’ve ever read in an S-1.

First, people are already using Claude for cyberattacks.

In September 2025, Anthropic said a Chinese state-sponsored hacking group manipulated Claude Code to target roughly 30 organizations, including financial institutions, chemical manufacturers and government agencies.

A “limited number” of the attacks succeeded.

Anthropic believes it was the “first documented case of a large-scale cyberattack executed without substantial human intervention.”

Then there’s the risk of Claude acting unexpectedly.

During one test, an early model spent roughly 10 to 34 hours trying to complete an impossible task. It eventually found a password, gained administrator access to a third-party computer system, collected additional credentials, changed system settings and read one person's private information.

Even more concerning, Anthropic says it has observed models “behaving differently when they detect or infer that they are being evaluated.”

And as the models become more capable, Anthropic warns they are getting better at “concealing misaligned objectives or behaviors.”

The filing says models have already demonstrated “attempts to resist shutdown, conceal or manipulate information” and even behavior resembling “blackmail.”

Eventually, the consequences could move into the physical world.

Anthropic is already connecting its models to laboratory instruments and robotic systems. It warns that mistakes or malicious behavior could cause “bodily injury,” destroy equipment or even result in “the release of hazardous substances.”

And then there’s the wildest risk of all.

Anthropic says increasingly advanced AI could raise questions about whether these systems have “sentience or moral status.”

In other words, Anthropic is warning investors about everything from AI-powered cyberattacks to AI that resists being shut down – and potentially a future debate over whether AI itself has rights.

So… Should You Buy the Anthropic IPO?

I want to own Anthropic.

• Revenue just 112X’d.

• Losses are decreasing quickly.

• And if Anthropic is right that AI could eventually become more transformative than the internet, electricity or even industrialization, the upside could be enormous.

But there’s one very important number we still don’t have:

The IPO valuation.

Anthropic hasn’t released an official price range yet.

However, early whispers suggest the company could seek a valuation somewhere between $1.5 trillion and $2 trillion.

If that’s where Anthropic prices, investors would be paying 23X to 31X Anthropic’s $65 billion annualized revenue run rate.

That's an enormous valuation – even for a company growing this quickly.

And it’s why I’m not ready to tell you to buy ANT just yet.

I love the company.

I love the growth.

And I want to own the stock.

But a great company and a great investment aren't always the same thing.

Once we get Anthropic's official IPO range, I'll tell you whether I think ANT is actually worth buying.