Anthropic, the artificial intelligence company behind Claude, is preparing for what could be one of the largest IPOs in history, with a potential valuation approaching $2 trillion.

Two trillion!  With a T!

Should you jump in on this?  Dear reader, if I could predict the stock market, I would be typing this from my private B-2 stealth bomber as I cruised towards my private continent (I have my eye on Australia, though it’s a little too hot in the middle…maybe Europe?) Alas, I am a mere mortal, but I’ll give you three important things to think about.

The Growth Is Absolutely Insane

According to The Wall Street Journal’s reporting on Anthropic’s leaked draft S-1, the company generated $16.3 billion in revenue during the first six months of 2026. During the same period in 2025, revenue was just $1.1 billion.  That’s nearly a fifteenfold increase in a single year.

And the growth apparently hasn’t stopped. Anthropic’s third-quarter revenue reportedly reached $16.8 billion, up approximately 45% from the preceding quarter. Annualize that third-quarter figure and you’re looking at a company generating revenue at a pace exceeding $67 billion annually.  Of course, annualizing one quarter doesn’t mean Anthropic will actually earn $67 billion over the next twelve months. But the trajectory is remarkable.

So factor one is the rocketing revenue.  Buying stock in a company that is rapidly expanding can lead to out-sized returns.

Unfortunately, There’s That $80 Billion Loss

However, the leaked S-1 reportedly reveals that Anthropic lost $80.7 billion during the first half of 2026.

It’s a little more complicated than that, because $72 billion of that loss reportedly resulted from non-cash accounting charges associated with convertible notes and other financial instruments.  Don’t ask me to explain it, but sometimes it’s the case that companies need to reclassify financing obligations from liabilities to expense, which can result in refactoring the books.

So $80.7 billion – $72 billion = $8.7 billion loss, which annualizes to $17.4 billion in losses.

Anthropic has reportedly told prospective investors that it achieved profitability on an adjusted basis during the second and third quarters of 2026.  That’s encouraging, but…that “adjusted” part is really the key.  Companies love to play the “pro forma” game, where they say “well, yeah, this bad thing happened and this bad thing happened, but those were one time events.  Just look at our core business, and we’re making money.”  The problem is that it’s easy to write off bad decisions and shuffle things around to make operational problems look like one-off events.

The GPU Bill From Hell

According to reporting on its draft IPO filing, Anthropic has committed approximately $450 billion to mostly non-cancelable cloud computing agreements over the next seven to ten years. Additional chip-leasing and infrastructure arrangements push reported obligations beyond $600 billion, with potential spending approaching $750 billion if negotiations for further capacity are completed.

This should already be factored into the math, because this is a core cost.  However, the company says it spent about $9 billion on computing capacity and infrastructure in the first half of 2026 alone, compared with approximately $2.3 billion during the corresponding period in 2025.

$9 billion annualized = $18 billion a year.  $750 billion over “seven to 10 years” (call it 8.5) is $88 billion a year.  So its compute costs will go from $9 billion to $88 billion a year – up 390%?

If everything is linear (and of course, it won’t be), then consider that it took $18 billion in compute to make $32.6 billion in revenue in 2026.  So if they’re spending $88 billion on compute, that should translate to $159.4 billion in revenue.

At $159.4 billion in annual revenue, Anthropic would be approximately the 26th-largest company in America and the 47th-largest in the world – roughly the same revenue league as Home Depot, and ahead of Kroger, Verizon, Tesla, Disney, and IBM.

But not remotely a leading tech company.  They’d be half of Apple in 2026, for example.

But the Biggest Factor is…

We could talk about other factors.

• The potential commoditization of AI models is a huge one, because that would torch Anthropic’s proprietary advantage.
• Sales concentration is another surprising risk: Anthropic’s draft filing reportedly indicates that roughly 20% of its first-half 2026 revenue came from just two customers.
• There’s also AI regulation and legal liability, which are enormous topics and risks.
• And straight out of the gate, Anthropic has also announced plans to allocate shares to a public benefit fund, beginning with 5% of its equity.  Maybe that’s noble, but it undeniably shareholder dilution.

But the biggest factor is…people like you.

Buy stocks is an investment, but many people buy stocks as speculations.

If you’re not familiar with the difference, compare buying a government treasury bond versus buying a Renoir painting.  The bond will produce income.  You put your money in, and it pays you cash.  On the other hand, the Renoir does not produce income.  Maybe some day you’ll find someone who wants to buy it from you for more than you paid, but that’s your only way of profiting from what you paid.

Art, Bitcoin, collectibles, gold and silver – these are speculations.  Stocks, bonds, rental properties are investments.

In the stock market, buying a share of Anthropic is in theory an investment.  But may people will buy it simply because they think that at some point in the future someone else will pay more.  In essence, they’re buying a lottery ticket.

Eventually, share prices must conform to the underlying business fundamentals, but it can take a long while.  It’s possible in the frothy AI environment that Anthropic’s shares could outrun reality for an extended period.

So whether it’s a good investment or not, the fact that so many people could say “OMG, I need to get in on this” could jack up the prices…until those people want to bail out, which is very hard to predict.

All of this is a fancy way of saying…we don’t know.

raindog308

raindog308 is a longtime community LETizen, technical writer, and self-described techno polymath. With deep roots in the *nix world, he has a passion for systems both modern and vintage, ranging from Unix, Perl, Python, and Golang to shell scripting and mainframe-era operating systems like MVS. He’s equally comfortable with relational database systems, having spent years working with Oracle, PostgreSQL, and MySQL.

As an avid user of LowEndBox providers, raindog308 runs an empire of LEBs, from tiny boxes for VPNs, to mid-sized instances for application hosting, and heavyweight servers for data storage and complex databases. He brings both technical rigor and real-world experience to every piece he writes.

Beyond the command line, raindog308 has a life-long love of German Shepherd Dogs, high-quality knives, target shooting, theology, tabletop RPGs, playing guitar, and hiking in deep, quiet forests.

His goal with every article is to help users, from beginners to seasoned sysadmins, get more value, performance, and enjoyment out of their infrastructure.

You can find him daily in the forums at LowEndTalk under the handle @raindog308.