This isn’t sustainable…

Anthropic’s draft IPO filing shows an $80.7 billion loss for the first half of 2026. Most of that is not what the headline makes it sound like, and the number that actually matters is somewhere else entirely.

We cover the economics of AI on this channel a lot, and this one is worth walking through carefully. Plenty of people talk about what the technology can do and how useful it is. Almost nobody talks about whether it is economically viable long term.

Start with the loss. About $72 billion of it is a non-cash accounting charge tied to convertible notes and other financing instruments the company issued to raise money. That is not somebody fuzzing up the books. Early investors hold notes that convert to shares later, and as the valuation climbs, those obligations get marked up. That is why the number balloons the way it does.

The other half of the story is that revenue is ahead of where people expected. $16.3 billion in the first half of 2026, against $1.1 billion in the same period last year. That is real growth.

Here is the part I actually want you to look at.

They spent close to $9 billion on compute and infrastructure in the first six months of this year. That is more than the $7.3 billion spent across all of 2025. Now look at what they are on the hook for. Roughly $450 billion in cloud commitments over the next decade, and the filing points to as much as $750 billion once deals still under negotiation close. Those commitments are noncancelable. They owe the money whether or not demand justifies it.

They have spent about $9 billion of it so far.

You as a consumer are not really the target market here, and even with prices on the higher tiers climbing, you are almost certainly paying less than what gets burned on your behalf. That is what a valuation like this buys you. Money is being thrown at this industry on speculation, not on economics that already work.

That is not sustainable. People hear AI bubble and think it means the technology goes away. It does not. The internet survived the dot com bubble. A lot of the companies did not, and some that made it spent the next decade climbing out of the hole. What is being spent now dwarfs what was being spent back then.

The filing itself raises the alarm. It says damages from a catastrophic AI incident could exceed what any insurance would cover. And roughly 86 percent of US venture money in the first half of this year went to AI. At some point somebody comes to collect.

What I worry about is what happens then, and whether these companies get propped up on a too big to fail argument. I would rather the chips fall where they may, even knowing that means a lot of people lose a lot of money.

Last thing, and this one is for the small business owners and anybody who wants to be one. This is a glimpse at how financing works at that level, and it is a completely different game from the one you are playing. Walk into a bank with a statement like this and ask for money and they will laugh you out of the building. So do not try to copy it. Just understand it is not the same game.